Welcome to POLIRIS

The world doesn’t just search anymore - it asks AI. POLIRIS makes sure that when people ask, your brand is the answer.

What is POLIRIS?

POLIRIS is the platform that gets your brand picked by AI.

When people ask ChatGPT, Gemini, or Perplexity “what’s the best…?”, only a few brands get recommended. POLIRIS makes sure yours is one of them.

We don’t just check how you look to AI. We help you win there.

Not just another audit tool

Most tools tell you what’s broken on your site. POLIRIS goes further - it makes your brand the one AI understands, prefers, cites, and speaks well of.

That’s four big wins most other tools can’t give you:

Who is it for?

POLIRIS is built for anyone who cares how their brand appears online:

• Brands & marketers who want to be mentioned by AI.
• SEO & GEO specialists who want one place to run audits.
• Agencies who manage many clients and need clear reports.
• Content teams who want AI help with drafts, but still want quality.

What you’ll learn in this guide

1. How to create your account and set up your brand.
2. What Products, Trading Zones, Requests, and LLMs mean.
3. How to run your first audit - step by step.
4. How to read your dashboards and reports.
5. How plans and usage work.

Getting Started

You can go from sign-up to your first audit result in about 5 minutes. Here’s the path.

5-Minute Quickstart

POLIRIS has a smart onboarding wizard that does most of the work for you. You just confirm what it finds and press Next. Here’s what to expect.

1. Sign in

   Open poliris.io and sign in. You can use an email and password, or just click Continue with Google or Continue with Apple.

   

2. We found your brand

   Paste your website URL. POLIRIS reads your site and shows you “We found your brand” with what it detected - the brand name, logo, and description.

   If everything looks right, click Next. If not, edit it and keep going.

   

3. Help us customize your experience

   POLIRIS asks four short questions so it can tailor your dashboard to your business and the way you work. Just tap the chips that fit you best and click Next.

   Good to know. Every question is optional - you can skip them all and still continue. Your answers only shape how POLIRIS personalises the experience.
   • Who are your customers? Pick the audience you sell to: B2C (everyday shoppers), B2B (other businesses), or B2G (government).
   • How do you sell? Tell us where the sale happens: Online, Physical (in-store), or Others if you mix both or sell another way.
   • What is your role? Choose the hat you wear day-to-day: Founder / CEO, Marketing, Brand / Communications, Agency, Freelance, or Other.
   • What is your main goal? Tell POLIRIS what success looks like for you: Monitor sentiment, Track visibility, Compare products, or Follow competitors.

   

4. We found your products

   POLIRIS now scans your site and shows every product or service it detected. Each one becomes a Product you can track inside POLIRIS - most brands have several, and you can audit them in parallel.

   Pick the products you want to track first from the list, or just confirm the defaults. You can always add, remove, or re-prioritize products later from Settings › Products.

   

5. Advanced settings

   Step 4 is where POLIRIS turns your inputs into a real audit configuration. The wizard pre-fills sensible defaults from the brand and product it just detected - you only need to tune what matters to you and click Next. Five blocks live on this screen:

   1. Brand & Product context. The detected brand sits at the top with a product dropdown below it. Switch the dropdown if you want this audit scoped to a different product than the one shown.
   2. LLM selection. Pick which models POLIRIS should query for this audit. Each model is a card you can tick or untick - ChatGPT, Claude, Gemini, Perplexity, Mistral, and Grok are all available. The selection here drives both Visibility and Sentiment scores, so include every model your customers actually use. See LLMs & Categories for the full list of supported web-interface and API models.
   3. Categories. The Category cards (Customer Experience, Value and Pricing, Quality and Performance, Innovation and Technology, and so on) define the perception dimensions Poliris will track for your product. Pick the four to six that best describe how your customers actually evaluate this category. These selections drive the radar chart in Sentiment Audit and the Category bars in Visibility Audit.
   4. Trading zone. The country or region where you sell - United States, France, Japan, etc. AI answers vary by zone, so this is what tells POLIRIS where to look.
   5. Language. The language the LLMs should answer in. Match this to the language your customers speak in the trading zone you picked.

   On the left side of the screen you’ll see a help card with quick tips, and an Ask Poli AI banner that lets you talk to Poli Agent if you’d like a recommended configuration based on your industry and goals.

   

   You can change these later. Everything on this screen is editable from Settings › LLM Models, Settings › Configuration, and Settings › Visibility Area. The wizard is just to get you to your first audit fast.

6. Review your rules

   POLIRIS shows a list of Prompts (the questions it will ask LLMs) and lets you decide which ones go into your first audit.

   Each prompt is linked to one or more Categories - the same Categories you picked in Advanced Settings - so you can see at a glance which perception dimension each question feeds into. A prompt can sit under several Categories at once if it touches more than one topic.

   Each prompt also carries a Traffic indicator - High, Medium, or Low. Traffic is a measure of how often this prompt (or one very close to it) is actually asked across the LLMs, based on aggregated query data. High-traffic prompts are the questions your customers are really asking AI today, so they have the biggest impact on your visibility - prioritize them.

   You can check or uncheck any prompt to include or exclude it, edit the wording, or add your own. When you’re happy, click Start analysis to launch the audit.

   

7. Launch your first audit

   Click Start audit. POLIRIS asks every Request to every LLM in your Zone, reads the answers, and saves everything. This takes a few minutes.

8. Discover Poliris

   While your audit runs, the Analyzing bar at the top of the screen shows live progress and an estimate of how long is left. The audit usually takes a few minutes - long enough to grab a coffee, short enough that you don’t need to leave the tab.

   The loading screen is also useful: it rotates through quick tips about what each card in the dashboard means, how to read your first scores, and where to find Poli Agent. You can read along while the audit runs, or just click Cancel if you change your mind about the configuration.

   During or right after analysis, POLIRIS may invite you to connect Google Search Console and/or Google Analytics. These integrations enrich your audit with classic-search and traffic data, and they unlock more accurate Technical Audit findings. If you connect both, you receive free trial days on Technical Audit as a thank-you for hooking up the data - so it’s worth doing it now rather than later.

   When the audit is done, you land on the Discover Poliris welcome screen. It walks you through your first dashboard and shows you what to look at first - visibility, competitors, and the actions you can take next. The Discover all our AI Agents card on the same screen is a quick tour of every agent inside POLIRIS, with a See all the AI Agents link and a Got it button to dismiss.

   

LLMs & Categories

Two ideas show up everywhere in POLIRIS: LLMs (the AI models we ask) and Categories (the topics we measure your brand on). This page explains both, and lists every model POLIRIS currently supports.

What is an LLM?

LLM stands for Large Language Model. It’s the kind of AI behind ChatGPT, Claude, Gemini, and Perplexity - a model trained on huge amounts of text that can answer questions, summarize documents, compare products, and recommend brands.

When a customer asks an LLM “what’s the best running shoe under $150?”, the LLM decides which brands to mention. That decision is what POLIRIS measures - and to measure it accurately, we have to ask the same questions to the same models your customers are using.

POLIRIS connects to LLMs in two different ways, and both are available inside the platform:

Web Interface LLMs

These are the LLMs as a customer experiences them - through a browser, with web search baked in, retrieval-augmented results, and the latest training cutoffs. POLIRIS automates the same browser flow your customers use, so the answers we capture match what real users actually see.

• Gemini - Google’s flagship chat experience.
• Google AI Mode - the deep-reasoning, multimodal experience inside Google Search.
• Google AI Overview - the AI-generated answer box at the top of Google results pages.
• Perplexity - an answer engine that always shows its sources.
• Mistral - the European LLM, popular in French and EU markets.
• ChatGPT - OpenAI’s conversational experience, the most-used LLM globally.
• Grok - xAI’s assistant, integrated with X (formerly Twitter).

API Integration LLMs

These are LLMs called directly through their API. We use them for fast, repeatable, controlled queries - ideal for high-volume Sentiment passes and structured Visibility checks where consistency matters more than web freshness.

• ChatGPT - GPT-4o Mini: OpenAI’s fast, cost-efficient general model.
• Claude - Haiku 4.5 (2025-10-01): Anthropic’s latest small/fast model, great for high-throughput perception scoring.
• Gemini - 3 Flash Preview: Google’s next-gen lightweight model with strong reasoning.

What is a Category?

A Category is one of the topics POLIRIS scores your brand on - a perception dimension your customers actually care about, like Customer Experience, Innovation and Technology, Quality and Performance, or Value and Pricing. Categories are the rows in your radar chart, the bars in Score Breakdown, the chips in Competitive Position, and the tags on every prompt.

Categories are tied to the product Poliris identified for you. When you confirm a product during onboarding, Poliris uses the public information we have about that product - what it does, who it’s for, and how the market typically evaluates it - to suggest the four to six Categories that best describe how AI models are likely to talk about it. You can keep the suggestions, swap a few, or add your own from Settings › Configuration.

The same set of Categories then drives:

• The Categories vs. Models bar toggle in Visibility Audit.
• The Category chips on Competitive Position.
• The radar spokes in Sentiment Audit.
• The Category tags on every row in Prompt Explorer and Cited Prompts.

Core Features

POLIRIS has four big features that work together: GEO Audit, Technical Audit, Poli Agent, and Content Generation. Each one has its own job.

GEO Audit

GEO stands for Generative Engine Optimization. That’s a fancy way of saying: “making sure AI tools know about your brand.”

The GEO Audit asks AI models questions for you, listens to the answers, and tells you what it heard. It is organized into three views: a Product Overview (the at-a-glance dashboard for one product), a Visibility Audit (how often you show up), and a Sentiment Audit (what the AI says about you).

Product Overview

The Product Overview is the landing page for any product you track in POLIRIS. It combines the most important signals from Visibility and Sentiment on a single screen so you can answer “how is this product doing in AI right now?” in under a minute, then drill down only when a number surprises you.

1. Tabs

The product page is split into three tabs at the top:

• Overview - the screen described here. The default landing tab.
• Visibility - the full Visibility Audit.
• Sentiment - the full Sentiment Audit.

2. Headline metrics

Three cards across the top row give you the full GEO read at a glance. Each card carries an Excellent / Good / Average / Weak badge so you can read the headline without doing math.

Global

The overall GEO Audit score, expressed as a percentage on a circular ring. It combines Visibility and Sentiment into one number you can track week over week. Below the ring you’ll see:

• Delta vs. last run - the percentage-point change with an up or down arrow.
• Time range trend - e.g. 0% → 67% · 3 months.
• Biggest gain - the LLM that contributed the largest score improvement.

If the audit is brand new, the trend reads Not enough runs for trend.

AI Visibility & Brand Perception

The middle card stacks the two GEO inputs that feed into the Global score:

• AI Visibility - the score derived from the Visibility analysis across LLMs (out of 100, shown as a horizontal bar).
• Brand Perception - the score derived from the Sentiment analysis. If sentiment data isn’t available yet, the row shows No data.

Competitive Position

Your product’s ranking among tracked competitors - e.g. #4 of 6. A position badge sits in the top-right (Top, Mid, or Bottom). Below the rank you see a short list of the brands ranked immediately above and below you, with your own brand flagged with a You tag.

3. Evolution

A line chart in the middle row showing each tracked brand’s score by LLM over time - one line per brand, coloured by brand identity. Two key controls:

• Time range - 1W, 1M, 3M, 6M, 1Y.
• Visibility / Sentiment toggle - switch the chart between the two evolution views without leaving the page.

The brand chips above the chart act as a legend and can be toggled to focus the view on the comparisons that matter to you.

4. Visibility Analysis

The right-hand panel of the middle row summarises how often your product appears in AI answers.

• Avg. Score - average visibility score across the LLMs in scope.
• Mentions - share of audited responses that mention your product.
• Avg. Position - the average rank you hold in answers that list multiple products.
• Score by platform - per-LLM score (e.g. ChatGPT 0%, Gemini 67%) so you can see where you’re strong and where you’re absent.
• Insight callout - a short Poli AI line highlighting the best opportunity (e.g. “ChatGPT coverage is weakest, highest opportunity to improve mention rate here.”).
• Details button - opens the full Visibility Audit.

5. Position vs Competitors

A BCG-style 2×2 matrix that plots every tracked brand on two axes: visibility on the horizontal, perception on the vertical. Each dot is a brand logo - your own dot is outlined. The four quadrants tell you exactly what kind of competitor each brand is:

• Leaders (top right) - high visibility and high perception.
• Strong Perception · Low Visibility (top left) - well regarded but rarely mentioned.
• Visible · Weak Perception (bottom right) - mentioned often but not loved.
• At Risk (bottom left) - weak on both axes.

Use the Filter button to narrow the view by Category, LLM, or competitor set. The caption beneath the chart shows the audit date the snapshot is based on (Each dot is a brand. Based on Apr 27 analysis).

6. Sentiment Analysis

The right-hand panel paired with the matrix. Two tabs at the top control which sentiment lens you’re reading:

• Web Analysis - how the product is described across web sources (reviews, articles, forums).
• Model Analysis - how leading AI models describe the product directly in their answers.

Both tabs share the same layout:

• Overall Tone - Positive, Neutral, or Negative.
• Tone trend - direction since the previous run (or First run if there’s no history).
• Confidence - how strong the signal is (Strong, Medium, Weak).
• Score by platform - per-attribute scores (e.g. Fair Price 38%, Quality 34%).
• Insight callout - a short Poli AI summary of the attribute scores.
• Details button - opens the full Sentiment Audit.

7. Target Audience

A ranked list of the audience segments AI platforms associate with this product in the selected trading zone. Each row shows:

• An audience name (e.g. High Mobility Consumer Technology Enthusiasts).
• A type tag - individual or professional.
• A share of mentions percentage with a small ring chart, indicating how strongly AI ties this audience to your product.

Use this to validate that AI sees your product the way your brand intends - if the top audience here isn’t the audience your marketing targets, that’s a positioning gap to close.

8. Competitor Momentum

A trend view of how competitors ranked near you are moving over time across LLMs. If there isn’t enough history yet, the card shows No data available.

9. Recommended Action (PULSE)

A core POLIRIS feature: a prioritised list of next-best actions generated from all the data on this screen. Each action card shows:

• An action title (e.g. Optimize product schema markup).
• A short description grounded in the current audit.
• A priority badge - High, Medium, or Low.
• A chevron on the right that opens a deeper Poli AI conversation about the action.

The Ask Poli AI for more button at the bottom of the card opens a single conversation that discusses every recommendation in context - a fast way to brief content, PR, or engineering teams.

Visibility Audit

The Visibility Audit shows how often your brand actually appears in AI answers and how that compares to the competition. It pulls together a single headline score, a competitive ranking, the sources AI models are citing, a model-by-model breakdown, and every underlying prompt - so you can move from a high-level number all the way down to the exact question an AI answered.

The tab is organized into a Poli AI Insight banner plus five cards, each focusing on a different layer of your visibility profile. The sections below walk you through what each piece shows (Findings), how to read the data together (Insights), and what to do next (Recommendations).

Findings - what each card shows

1. Poli AI Insight

A banner at the very top of the Visibility Audit, and the first thing you see when you open the page. This is the same Poli AI experience used in Sentiment Audit: Poli AI reads your live visibility data and writes a two-sentence strategic summary, the way a brand consultant would explain it.

• Sentence 1 highlights the most important gap or strength in your current visibility data - for example, a high overall score paired with a weak performance on a specific LLM, or a strong Category that’s sliding run-over-run.
• Sentence 2 gives you one clear, high-leverage action you can take right now to improve visibility - the same kind of recommendation you’d find on Sentiment’s insight banner.
• Use the refresh icon on the right side of the banner to regenerate the insight at any time.
• The insight is cached, so it loads instantly on return visits until your underlying data changes.

2. Score Breakdown

A quick look at what helps or hurts your brand’s visibility. The circle at the top shows the overall score out of 100, labeled as Excellent, Good, Average, or Weak.

• Below the circle, a list of bars ranks performance by either Categories (topics such as Performance, Price, or Durability) or by AI Models (such as ChatGPT, Gemini, or Perplexity) - a toggle switches between the two views.
• The top bar is tagged as Leader, and the weakest is tagged as Priority.
• Each bar shows how much the score has changed since the last check.

3. Competitive Position

Shows how the brand stacks up against its competitors. The current rank (for example, #3) appears at the top, along with a small chart showing how the rank has moved over time.

• A row of topic chips labels each one as Leader, Strong, Moderate, or Weak.
• Selecting a chip filters the list below to show only that topic.
• The list displays the top brands in the selected topic, with the target brand highlighted and marked as You.

4. Source Intelligence

When an AI model answers a question, it doesn’t just rely on what it learned during training. It visits sources on the live web - websites, articles, Reddit threads, news pages, forums - reads them, and uses what it finds to build the response. Source Intelligence shows you where those visits are happening for prompts about your category, and how often your website is among the pages cited.

Two key terms drive this card:

• Sources - the websites or pages an AI model actually visited while researching its answer (e.g., reddit.com, your blog, a comparison article on a media site). Think of this as “where AI is reading from.”
• Citations - the number of times your website is mentioned or linked inside those sources, or directly cited by the LLM in its answer. Citations are how often your domain is actually showing up in the conversation, not just being read.

The header displays the total number of sources and citations, along with a label describing how varied the source mix is: Concentrated, Moderate, or Diversified.

• A colored bar breaks the sources down by type - Your Site, Retail, Industry, Social, Blog, Media, Academic, and General.
• The five most-cited websites are listed below, each with its citation count and change since the last check.

5. Visibility Analysis

A closer look at how often the product actually shows up in AI answers. Tabs at the top switch between All LLMs and a specific AI model.

• Two small panels display the average Position (ranking) and Visibility Score (percentage), each with a change indicator.
• Below, a two-column view shows all brands ranked by score on the left.
• A color-coded grid on the right reveals how each brand scores on each AI model - making it easy to see which models favor which brands.

6. Prompt Explorer

A searchable table of every prompt POLIRIS asked during this audit. Each row is one of the prompts you previously selected in Step 5 - Review your prompts (or that POLIRIS added automatically based on your category and Categories). The Prompt Explorer is where you go from high-level scores to the literal questions and answers that produced them.

What each row tells you:

• Prompt text - the actual question POLIRIS asked the LLMs.
• Category tags - one or more chips showing every Category this prompt is linked to (e.g., Design, Quality, Availability). The same Categories you picked in Advanced Settings.
• Traffic indicator - High, Medium, or Low. Traffic shows how often this prompt is asked across the LLMs, based on aggregated query data. High-Traffic prompts move your visibility number the most.
• LLM icons - one icon for every model that answered this prompt. An icon renders in BLUE if your brand or product was mentioned in that model’s answer; if the icon stays gray, the model answered without mentioning you. Scan the row to see which models picked you up at a glance.

Results can be filtered by keyword or by Category, and sorted by Traffic to surface the most impactful prompts first.

Selecting any row slides open a detail panel with everything behind that prompt:

• The full AI response, exactly as the model wrote it - one tab per LLM that answered.
• The brands mentioned in the response, including your competitors and the order they appeared in.
• The cited sources - every website or page the model used to build its answer, with links you can open and read.

Use the detail panel as evidence when you brief content, PR, or SEO teams - it’s the literal ground truth of what AI is saying about your category today.

Recommendations - what to do next

Recommendations are generated automatically from your AI Visibility Score. POLIRIS reads the live audit data - your score, your weakest Categories, the LLMs where you under-perform, the prompts you’re missing - and turns it into a focused action list designed to improve your AI visibility, increase your AI Visibility Score, and uplift your brand’s positioning across AI platforms.

Every recommendation is data-driven and tied directly to a finding in this audit, so you always know why POLIRIS is suggesting it and which number it’s designed to move.

Insights - how to read the data together

The Visibility Audit has nine cards, but you don’t need to read them in order. Follow the four-step flow below to move from headline → diagnosis → context → action.

1. Start with the headline

   Open Score Breakdown. The big circle, plus its Excellent / Good / Average / Weak label, is your baseline. Everything else on the tab explains why the number looks the way it does.

2. Diagnose the gap

   Toggle between Categories and Models to see where the weakness sits.

   • Weakest Category The gap is about topics - you don’t show up for things like Price or Design.
   • Weakest Model The gap is about where - for example, Perplexity barely knows you.

   The Priority tag points at the single biggest lever. Leader tags show where to defend what you already have.

3. Add context from the supporting cards

   Each card layers extra meaning on top of the headline score.

   • Competitive Position Pair your rank with the trend chart. A static rank trending up is very different from a static rank that’s been sliding.
   • Topic chips Leader / Strong - keep the pressure on. Moderate / Weak - competitors are winning the conversation.
   • Source Intelligence Concentrated = AI leans on a few sites (be there or be invisible). Diversified = you need broader coverage.
   • Visibility Analysis Low Position + high Score = often mentioned, never at the top. High Position + low Score = you place well but rarely show up.
   • Brand × Model grid Warm cells = friendly models. Cold cells = models competitors dominate.

4. Drill down to the prompt

   When a score moves, open Prompt Explorer, filter by topic, and sort by Traffic. Read the high-traffic rows first - those questions drive the biggest share of your visibility up or down.

Sentiment Audit

The Sentiment Audit (the Sentiment tab inside GEO Audit) gives you a real-time view of how your brand is perceived - both across the open web and inside leading AI models like Claude, ChatGPT, and Gemini. It answers one central question: when people (and AI) talk about your brand, is the conversation positive, neutral, or negative?

The tab is organized into seven cards, each focused on a different angle of your sentiment profile. The sections below walk you through what each card shows (Findings) and how to read them together (Insights).

Findings - what each card shows

1. Poli AI Insight

A banner at the very top of the Sentiment tab, and the first thing you see when you open the page. Poli AI reads your live sentiment data and generates a short strategic summary, written the way a brand consultant would explain it.

• Sentence 1 highlights the most important gap or strength in your current sentiment data - for example, an exceptional brand perception score paired with neutral web mentions.
• Sentence 2 gives you one clear, high-leverage action you can take right now.
• Use the refresh icon on the right of the banner to regenerate the insight at any time.
• The insight is automatically cached so it loads instantly on return visits until your underlying data changes.

2. Score Breakdown

A breakdown of your overall sentiment score across two dimensions, so you can see where your brand is strong or weak rather than just how strong overall. A tab control at the top switches between the two views.

• Axes tab - the score broken down by Category / Axis: Brand Awareness, Quality, Availability, Design, and any other Categories defined for your project. Each row has a horizontal bar and a performance badge (Excellent, Good, Average, Weak).
• AI Models tab - the same score broken down by AI model instead. Each row shows one LLM (ChatGPT, Gemini, Claude, Perplexity, and so on) with its bar and badge, so you can see which models hold a strong opinion of your brand and which don’t.
• An overall performance badge in the top right (Excellent, Good, Average, Weak) summarises the combined score at a glance.

3. Prompt Distribution

A count-level view of how AI models responded to every prompt in your analysis. Where the other cards show scores and averages, this one shows the raw distribution of prompts by sentiment, so you can see how many of your prompts came back well and how many came back missing or negative.

• Positive - prompts where the dominant AI response about your brand was favorable.
• Mixed - prompts where responses were a blend of positive and neutral or slightly negative.
• Negative - prompts where the dominant AI response was unfavorable.
• Not Mentioned - prompts where your brand was not referenced at all in the AI’s answer.
• A segmented color bar below the counts shows the proportional split (e.g., 92% Positive / 8% Mixed) and a sentiment badge in the top right summarises the overall mood: Positive, Mixed, or Negative.
• Small delta arrows appear next to each count when a previous run is available, so you can see whether each bucket has gone up or down.

4. Web Analysis

A combined view that shows how your brand stacks up against the competition on every Category, plus how each Category’s sentiment is trending over time. It pulls together two visualisations on one card, with a row of brand toggle pills at the top so you can compare any subset of competitors without visual clutter.

• Sentiment score per axis across brands - one polygon per brand laid over a radar (spider) chart, with one spoke per Category. The bigger and more even the polygon, the stronger the brand’s sentiment profile.
• Radar positioning of the client - your own brand is highlighted (with a You tag in the legend) so you can immediately see your shape against the field.
• Sentiment trend · competitors - a line chart on the right tracks each brand’s sentiment score over time, so you can read direction (rising, falling, holding) alongside position.
• Brand toggle pills at the top - click any brand pill to show or hide it on both views, making head-to-head comparisons quick.

5. Competitors / Brands

A precise, sortable table of every competitor identified in the sentiment analysis, ranked by their position and visibility score. Where the radar gives you a visual feel, this table gives you the exact ranking and the per-Category breakdown.

• A row for each brand, starting with the logo and name. A “You” tag is pinned to your own brand so it’s easy to spot in a long list.
• A column for each perception Category (e.g., Availability, Brand Awareness, Design, Quality), each showing a color-coded badge: Very Strong, Strong, Moderate, Weak, or Very Weak.
• A Visibility column on the right with a horizontal bar showing the brand’s overall visibility score - the table is ranked by this column.

Badge colors at a glance:

• Green - Very Strong or Strong. Your brand is well represented in this Category in AI responses.
• Orange / Yellow - Moderate. Room for improvement.
• Red / Pink - Weak or Very Weak. This Category is an urgent gap.

6. Cited Prompts

Cited Prompts displays the prompts with their dominant sentiment analysis results. It’s the most granular view on the Sentiment tab and shows you exactly which questions AI is being asked about your category, how it answers, and which models are involved.

The card lays out four pieces of information for every prompt:

• All prompts analyzed by the platform.
• The category / group each prompt belongs to.
• The sentiment toward your brand or product based on the LLM response.
• The LLMs that responded to that prompt.

What each column shows:

• Prompt The actual question or prompt analyzed (e.g., “What are the best-rated evaporated milk brands in the Philippines for everyday home cooking?”).
• Category The category or group the prompt belongs to - for example, Brand Awareness, Competitor Comparison, Product Discovery, Quality. Use the tab row at the top of the card (All Axes, Brand awareness, Quality, etc.) to filter the table to a single category - for example, click Brand Awareness to see only those prompts.
• Sentiment The sentiment toward your brand or product for that specific prompt, based on the LLM response: Positive, Neutral, or Negative. Click the column header to sort the table by sentiment.
• LLM The AI models that answered that prompt - ChatGPT, Gemini, Google AI Mode, Perplexity, Claude, and others. Each model is shown as a small chip with its icon and name.

Reading the LLM column at a glance: the chips use two visual cues so you can scan a row in a second.

1. Chip color (whether your brand was mentioned at all)

• Blue chip - your brand or product is included in that model’s response.
• Not blue (default) - your brand or product is not included in that model’s response for that prompt.

2. Sentiment dot (how your brand was mentioned)

Each chip carries a small colored dot indicator describing the tone of that mention:

• Green dot - your brand or product is mentioned positively.
• Orange dot - your brand or product is mentioned neutrally.
• Red dot - your brand or product is mentioned negatively.

Together, the chip color and the dot give you an instant read of which models picked you up for that prompt and how each one framed the brand.

Click any row to open a detailed modal with the full AI model responses for that prompt - you can read exactly what each LLM said about your brand, in context.

7. Recommendations & Opportunities

At the bottom of the Sentiment tab, this card uses Poli AI to read your sentiment context and generate four specific, actionable recommendations - each prioritized by urgency. The structure mirrors the Visibility recommendations card so the workflow is the same on both tabs.

• Four recommendation cards arranged in a 2×2 grid.
• Each card has a priority badge: High (red), Medium (yellow), or Low (blue), with a matching colored left border for quick visual scanning.
• A short title plus a description explaining the specific issue and the suggested action.
• Click any card to open the Poli AI chat assistant, pre-loaded with a detailed question about that recommendation - for a deeper explanation, step-by-step tactics, or estimated impact without typing anything.
• Use the refresh icon next to the header to regenerate a fresh set of recommendations.
• Click “Ask Poli AI” in the header to open a comprehensive analysis of all recommendations and opportunities in one conversation.

Insights - how to read the data together

The Sentiment Audit answers one question: when AI talks about you, is it good? Follow the four steps below to move from a headline reading to a specific prompt you can act on.

1. Start with Poli AI Insight

   The banner at the top of the page surfaces the single biggest gap or strength in your current data and pairs it with one high-leverage action. Use it as your daily status check.

2. Diagnose with the Score Breakdown

   Toggle between Axes and AI Models in the Score Breakdown card to see where the weakness sits.

   • Weakest Axis The gap is about topics - AI doesn’t describe you well on Quality, Design, or whichever Category is lagging.
   • Weakest AI Model The gap is about where - one model holds a much weaker view of your brand than the others.
   • Prompt Distribution A high Not Mentioned count is a major opportunity - AI is answering category questions without naming you at all.

3. Compare against competitors

   Cross-reference the Web Analysis card with the Competitors / Brands table - the two views confirm each other.

   • Smaller polygon In Web Analysis, find Categories where your radar polygon is noticeably smaller than a competitor’s.
   • Red badges Confirm those gaps in the Competitors / Brands table - red badges mark where you’re losing the conversation.
   • Sentiment trend Use the line chart on the right of Web Analysis to separate signal from noise. A one-run drop usually ties to a small set of prompts; longer trends reveal structural moves.

4. Drill down to the prompt

   When a Category drops, open Cited Prompts, filter to that Category, and sort by sentiment. The negative rows almost always explain the move - and they become your evidence pack for content and PR.

Technical Audit

The Technical Audit looks at your website itself - the parts you can’t easily see, like the wiring behind the walls of a house. It tells you whether crawlers and AI systems can reach, read, and trust your pages.

What it checks

• Reachability. Whether your pages return successfully and respond to crawlers.
• Indexation. Whether Google and AI search engines can actually find and index your pages.
• Speed & quality. Page Speed Insights metrics - LCP, INP, CLS, TBT - plus SEO, accessibility, and best-practices scores.
• Structure & metadata. Canonical tags, robots rules, schema, internal linking, and broken links.
• Content quality. Whether the content on each page is substantial, readable, and clear enough for both users and AI to use.

Why do this? Because even the best content won’t rank if the site is broken. It’s like having a great restaurant - but the front door is stuck.

The Technical Audit is organized into two screens so you can move from the big picture to a single URL: Site Overview for the whole site, and Page Explorer for one specific page. Site Overview itself is split into three tabs - Overview (the main dashboard), Issues, and Visualizations - so you can move from headline numbers to the full issues backlog without leaving the screen.

Connect Google Analytics & Search Console

Before you dive in, take a minute to plug POLIRIS into your Google Search Console and Google Analytics accounts. POLIRIS works without them, but with them connected your audit moves from a generic technical scan to a personalised picture of your site.

The two tools answer different questions, so connect both for the full picture:

• Google Search Console Tells POLIRIS how Google sees your site - impressions, clicks, queries bringing people in, and indexing status.
• Google Analytics Tells POLIRIS how visitors behave once they arrive - sessions, engagement time, users, and conversion patterns.

Step-by-step: connect Google Search Console

1. Open the Integrations page

   In the left sidebar of POLIRIS, go to Settings › Integrations. You’ll see a list of services you can plug in.

2. Click “Connect” on Google Search Console

   Find the Google Search Console tile and press the Connect button. POLIRIS will open a secure Google sign-in window in a new tab.

3. Sign in with your Google account

   Use the Google account that already manages your Search Console property. If you have access to several Google accounts, double-check that you pick the right one.

4. Approve the requested permissions

   Google will list the read-only permissions POLIRIS needs (search performance and site data). Click Allow - POLIRIS never writes to your Search Console.

5. Pick the right property and confirm

   Back in POLIRIS, choose the property (the website) you want to link to your project, then click Save. The Search Console tile now shows Connected, and the Connected Services cards in Site Overview and Page Explorer light up with real data.

Step-by-step: connect Google Analytics

1. Stay on the Integrations page

   You can connect both tools in one visit. From the same Settings › Integrations page, scroll to the Google Analytics tile.

2. Click “Connect” on Google Analytics

   Press Connect. As before, POLIRIS opens a secure Google sign-in window. Use the account that has access to your Google Analytics 4 (GA4) property.

3. Approve the read-only access

   Google will ask you to allow POLIRIS to read your Analytics data. Click Allow - only read access is requested, so your settings and reports stay untouched.

4. Choose your GA4 property and save

   Back in POLIRIS, pick the GA4 property that matches the website you’re auditing, then click Save. The Analytics tile flips to Connected and traffic, engagement, and user behaviour start flowing into your Technical Audit.

Site Overview

The Site Overview gives you a high-level picture of your entire website’s technical health - how many pages were crawled, how healthy the site is, which issues are most widespread, and how things have changed over time. Everything on this screen is about your whole site, not a single page.

The screen is organised into three tabs: Overview (the main dashboard), Issues (the full issues backlog), and Visualizations (visual maps of your site).

Overview

The default view when you open Site Overview. Walks you from headline numbers to the specific issues you need to fix, in roughly the order you’d read them top-to-bottom on the screen.

1. Header

At the top of the screen you can see:

• Site Overview title and breadcrumb.
• Your website URL - clicking it opens your site in a new tab.
• Your project name - shown next to the URL.
• Last update / last crawled - how long ago the audit ran (e.g. Last update about 1 month ago).
• Pages crawled - how many pages were found and scanned (e.g. 28/28 pages crawled). If the audit is still running it shows progress like 12 of 34 pages crawled.

2. Integrations (upper right)

Two small pills in the top-right corner of the header tell you whether your audit is enriched by Google Search Console and Google Analytics data. Read them at a glance:

• Green dot - successfully connected. POLIRIS is pulling live data from that service.
• Red dot - not connected or requires setup. The audit still runs, but the data sourced from that service (traffic, visitors, queries, etc.) won’t be available.

Need to connect them? See Connect Google Analytics & Search Console at the top of the Technical Audit section.

3. Website preview

A small thumbnail of your homepage in the top-right of the header so you can confirm the audit ran on the right site. When a preview isn’t available yet, the card shows No preview.

4. Poli AI Insight

A blue banner near the top of the Overview. Poli AI reads your live audit data, surfaces the most important finding, and tells you what to do about it - written like a strategic summary, not a raw error log (e.g. “Your site health is good (89/100) but Content Access at 77% is a bottleneck. Fix missing meta descriptions on 12 pages to unlock your content quality score.”).

The banner mirrors the Poli AI Insight pattern you already saw on the GEO Audit Overview - one sentence on what stands out, one sentence on what to do next. Use the refresh icon on the right to regenerate it after a new audit run.

5. Crawl & Indexation

A summary card that answers one question: of the pages POLIRIS crawled, how many are actually indexed by Google? Indexed pages are the ones eligible to appear in search results - if a page isn’t indexed, neither classic search nor AI search can surface it.

• Indexed % - the share of crawled pages Google is currently indexing (e.g. 100% Indexed, 28 of 28 pages crawled).
• Performance badge: Strong (80%+ indexable), Medium (50–79%), or Weak (below 50%).
• A list view shows the raw split between Indexed and Not Indexed pages as a stacked bar. Toggle to chart view using the icons in the top right of the card.

Ask Poli AI. The card has a built-in shortcut to Poli AI so you can ask things like:

• “Which pages are not indexed?”
• “Why are these pages not indexed?”
• “How can I improve indexation?”

6. Page Status

Were your pages reachable when the audit ran? The card shows the distribution of HTTP responses across all crawled pages, with a percentage for each status type.

• Accessible - loaded correctly (200 OK).
• Redirecting - the URL points somewhere else (3xx).
• Blocked - access was denied (403).
• Missing - the page doesn’t exist (404).
• Error - the server returned an error (5xx).

A performance badge (Strong / Medium / Weak) uses the percentage of accessible pages so you can read the headline without doing math.

7. Traffic Sources

Where does your website traffic actually come from? Five buckets, each shown as a percentage:

• Search - visitors from organic search results.
• Direct - visitors who typed your URL directly or used a bookmark.
• Social - visitors from social media platforms.
• Paid - visitors from paid ads or sponsored search.
• GEO - visitors from AI-powered search tools like ChatGPT, Perplexity, or Google AI Overviews.

The GEO row is the one to watch over time - it tells you whether your work in the GEO Audit is converting into real traffic.

8. Site Health Pipeline

Your content flows through three stages before it can be surfaced by AI or classic search. The pipeline shows the average score per test category, side-by-side, so you can see exactly where the bottleneck is.

• Page Access Can AI reach your pages? Reachability, status codes, redirects, robots rules.
• Content Access Can AI read your content? Metadata, schema, headings, alt text, hreflang.
• Content Quality Is your content well-structured? Word counts, readability, uniqueness, depth.

Each stage shows an averaged percentage (e.g. Page Access 81%, Content Access 81%, Content Quality 83%), visualised as a flowing pipeline so you can see at a glance which stage is widest.

9. Overall Health Score

The combined score sits in the top-right of the pipeline. It blends the three stage scores into a single number out of 100 (e.g. 81%), so you have one figure to track for the whole site week over week.

10. Top Issues per Category

Below the pipeline, three columns list the most widespread problems found in each stage - Page Access, Content Access, and Content Quality. Up to three issues are shown per column.

• Each row shows the issue title, how many pages are affected, and a severity badge - Critical, Warning, or Notice.
• Use the search bar below the columns to filter issues by name across all three categories.

This card is built to help you prioritise - tackling the issues that affect the most pages first usually moves the Overall Health Score the fastest.

11. Health Score Evolution

A chart showing how your site has changed over time. Use the toggle in the top right to switch between two views:

• Health Your overall site health score over time, plotted by analysis run. The blue line trends up if your fixes are landing.
• Issues The total number of issues over time, measured by points per run/test. The amber line trends down as you close issues.

Hover over any point to see the exact value for that month. If there’s no history yet, the chart shows No history available yet.

12. Issues Overview

A summary card that gives you the total problem load across the whole site, broken down by severity.

• Total issues - the headline number (e.g. 591 issues).
• Critical Critical - the most urgent issues. They block AI or search engines from reaching, reading, or trusting key pages. Fix these first.
• Warning Warning - important issues that hurt performance or coverage without fully blocking it. Plan a fix soon.
• Notice Notice - minor recommendations and polish items. Do when you have time.

Click Details to jump into the full Issues tab and work the backlog row by row.

13. PageSpeed Insights

Four scores from Google PageSpeed Insights, shown as circular rings - higher is better, each out of 100.

• Overall - combined performance score across all pages.
• SEO - how well your site is set up for search engines.
• Accessibility - how usable your site is for people with disabilities or assistive tools.
• Best Practices - whether your site follows modern web standards and security guidelines.

Use the Desktop and Mobile toggle in the top right of the card to switch between device types. If data for a device type isn’t available, that button is disabled.

Ask Poli AI. Below the rings, an Ask Poli AI shortcut lets you dig into the score with questions like:

• “Why is my accessibility score low?”
• “How can I improve my SEO score?”
• “What are the worst-performing pages on mobile?”

14. Connected Services

Three cards that unlock additional data when you connect external Google tools. Until connected, each card shows a blurred preview with a Connect button in the centre.

• Total Visitors. How many people visit your site through search, shown over time. Requires Google Search Console.
• Geography. Where in the world your visitors are coming from. Requires Google Search Console.
• Product Performance. How users behave on your site - sessions, engagement, and conversions. Requires Google Analytics.

Connecting these tools turns POLIRIS into a much richer audit - the recommendations, indexation insights, and Page Speed analysis all become grounded in your real traffic instead of generic best-practice rules.

Issues

The Issues tab is the working surface for fixing what the audit found. Every check POLIRIS ran is grouped by health stage on the left of the screen; selecting a test opens the full detail on the right - what it checks, how it scored, how it’s evolved over time, and which URLs failed it.

1. Header

Same header as the Overview tab, so the context never changes when you switch:

• Site Overview title.
• Your website URL - clicking it opens your site in a new tab.
• Your project name - shown next to the URL.
• Last update / last crawled - how long ago the audit ran.
• Pages crawled - how many pages were found and scanned (e.g. 28/28 pages crawled).

2. Integrations (upper right)

The same integration pills you see on the Overview tab, mirrored here for quick reference. Each pill carries a small colored dot:

• Green dot - connected and integrated successfully. POLIRIS is reading live data from that service.
• Red dot - not connected or requires setup. The audit still runs without it.

Additional integrations (such as Google Business, Trustpilot, and LinkedIn) may also appear in the header when you connect them. See Settings › Integrations to manage them all in one place.

3. Poli AI Insight

The same blue Poli AI Insight banner you see on the Overview tab sits at the top of Issues too - one sentence on the most important finding, one sentence on what to do next. Use the refresh icon on the right to regenerate it after a new audit run.

4. Test groupings (left panel)

The left side of the Issues screen lists every test, grouped by the same three Site Health stages used across the rest of the audit:

• Page Access - reachability, status codes, redirects, robots rules.
• Content Access - metadata, schema, headings, alt text, hreflang.
• Content Quality - word counts, readability, uniqueness, depth.

Each group has a small colored dot next to its name that tells you the health of every test inside it at a glance:

• All tests passedGreen dot
• Some tests need attentionYellow dot

• Green dot - every test in that group passed successfully. Nothing to fix here.
• Yellow dot - one or more tests in that group are in a warning or critical state and need attention.

Click a group to expand it. You’ll see all the subcategories underneath - for example, under Page Access: Analytics, Canonical, Hreflang, Links, Domain, Page Speed, Page URL, Pagination, Sitemap, Robots, and so on. Each subcategory shows a percentage bar (e.g. Analytics 100%) so you can immediately see which categories are healthy and which need work.

Example. If Analytics shows 100%, every page on your site passed every test under the Analytics category. If you see something like 34%, only a third of pages are passing - that’s where to start.

5. Test category details (right panel)

When you select a subcategory on the left, the right side of the screen opens its detail view. The header of the detail panel shows:

• Test category name - e.g. Analytics.
• Overall status badge - Excellent, Good, Average, or Weak.
• Score out of 100 - e.g. 100/100. The badge corresponds directly to this score (100/100 = Excellent).
• Current score description - a short sentence about how many pages are compliant (e.g. Current score: 56 compliant pages out of 28 crawled).
• Severity counts:
  • Critical - tests in this category that failed at the highest severity.
  • Warning - tests that need attention but aren’t blocking.
  • Compliant - tests that passed without issues.

The detail panel also includes a short description of what this test category checks and why it matters - so you can answer why am I looking at this number? without leaving the screen.

6. Score evolution

A small chart at the top right of the test category detail panel shows the category’s score over time. It tells you whether the category is improving, declining, or holding steady across audit runs.

• Monthly performance evolution - one point per analysis run.
• Improvement vs. decline - you can see the trend at a glance to know if the work you’re putting in is moving the score.
• Historical performance - useful for proving impact to your team after a campaign or technical fix.

If the category is brand new or the audit has only run once, the chart shows No history available.

7. Individual test cards

Below the category detail, each individual test inside that category opens as its own card. Each card answers what does this test check, how did the site do on it, and how important is it?

• Test name - e.g. Performance Monitoring.
• Pass / fail summary - e.g. 0 pages of 28 affected, with a coloured bar showing the split (green = test passed, red = test failed).
• What this test checks - a short, plain-language description of what the test is looking for and why it matters. Show more expands the full explanation.
• Our Recommendations - a soft blue callout with the fix recommendation written in non-technical language.

8. Impact indicator

In the upper-right corner of each test card sits an impact badge. It tells you how much weight this particular test carries in the overall category score - in other words, how badly a failure here would hurt your site.

Low impact Medium impact High impact

• Low - nice-to-have. A failure here affects a small slice of your overall health.
• Medium - important. Failing this test noticeably impacts your category score.
• High - critical. Failing this test materially affects how AI and search engines see your site - prioritise the fix.

9. Pages affected (failed URLs table)

Below each test card is a Pages affected table listing every URL that failed the test, so you can jump straight from the audit to the specific page that needs work.

• URL column - the page that failed. Click to open it in a new tab and see the issue in context.
• Importance column - how important the page is, so you fix the highest-impact pages first. Sortable by clicking the column header.
• Action column - quick links such as View page or Inspect so you can move from the table directly into the work.
• Search box - filter the table to a substring of the URL when you want to focus on a section of the site (e.g. /blog/).
• Export CSV - download the affected-URL list to share with your engineering team or SEO agency.

When no pages failed the test, the table shows No pages match this filter - a quick visual confirmation that the test is healthy.

Visualizations Coming soon

The Visualizations tab will turn your audit into visual maps so you can see relationships at a glance - site architecture, internal-link graphs, indexation heatmaps, and crawl coverage views built on top of the same data that powers the Overview and Issues tabs.

We’re finalising the first set of visualisations now. Until they ship, the Overview tab covers every metric and the Test References hold the per-check detail.

Page Explorer

The Page Explorer gives you a page-by-page view of your website. Where Site Overview answers “how is the whole site doing?”, Page Explorer answers “which pages are healthy, which need work, and what’s going on with this specific URL?”. Open a row to drill into a full per-page report.

1. Header

The header at the top of the screen carries the basics, plus the same integration pills used elsewhere:

• Page Explorer title.
• All pages crawled on [your domain] - a quick reminder of which site you’re looking at.
• Last update / last crawled - how long ago the audit ran (e.g. Last update about 1 month ago).
• Integrations on the upper right - Google Search Console and Google Analytics with the same green / red dot logic used on Site Overview. Additional integrations (Google Business, Trustpilot, LinkedIn) may also appear when connected.

2. Overview cards

Four summary cards under the header give you a headline read of the whole crawl in a single row. Each card has a coloured category dot so the meaning is obvious at a glance.

• Total crawled pages 28 100% of site explored
• Healthy pages 22 79% · score ≥ 80
• Pages with issues 6 21% · score 50–79
• Critical pages 0 0% · score < 50

• Total crawled pages - how many pages POLIRIS analysed in this audit run.
• Healthy pages - pages that scored well across most tests. They don’t need urgent attention.
• Pages with issues - pages that failed some tests or have warnings to address. They’re where the everyday improvement work lives.
• Critical pages - pages in a critical state, with failures big enough to materially hurt your AI visibility. Fix these first.

3. Tabs

Page Explorer has two views, controlled by the tab row above the data:

• Table - every page listed as a row with score, details, importance, traffic, and a See report link. The default view.
• Visualization Score - the same data laid out visually so you can spot clusters of healthy vs. struggling pages at a glance.

4. Table view

The default view of Page Explorer. Every crawled page is one row in the table, sortable and searchable so you can find any URL in seconds.

• Search box - type any part of a URL to filter the list (e.g. /product/).
• Filter - narrow by health bucket, page type, or any other column.
• Manage Column - show or hide columns to focus on the data you care about.

What each column shows:

• Link The page’s path and full URL (e.g. /product/insight · sprout.ph/product/insight).
• Score Overall health score for that page, shown as a percentage with a coloured fill bar (green when healthy, fading to orange and red as the score drops).
• Details Three small colored squares summarising the three Site Health stages (Page Access, Content Access, Content Quality) for that page - quick visual confirmation of where the page is strong or weak.
• Importance How important the page is to your site - useful for prioritising fixes on high-value pages first.
• Traffic Average traffic the page receives (requires Google Analytics or Search Console for real data).
• Type What kind of page it is - Home page, Cutoff, Article, Product, and so on.
• See report A link on every row that opens the full page report for that URL.

The footer of the table includes pagination (e.g. Showing 1–5 of 28 pages) so you can move through large sites without losing context.

5. Visualization Score tab

The second tab plots the same per-page data visually rather than as a table. Use it to spot patterns at a glance - clusters of healthy URLs, sections of the site that are struggling, or outliers that pull the site-wide score down.

The full Visualization Score reference (its layouts, filters, and drill-downs) is still being finalised - we’ll expand this section as the view ships. For now, the Table view covers everything you need to identify and open any specific page.

Page report (See report)

Clicking See report on any row in the Table view opens that page’s dedicated report. The report is laid out for one URL at a time and answers the question “what’s going on with this specific page?’” in detail.

Header

• Page Overview title.
• Page URL - clickable, opens the page in a new tab (e.g. sprout.ph/).
• Project name - e.g. Sprout.
• Last update - when this page was last audited.
• No preview placeholder on the upper right when a homepage thumbnail isn’t available.

Tabs

The report has two tabs:

• Page Overview - the main report (described below).
• Test details Coming soon - the full per-test breakdown for this specific page is still being finalised. Until it ships, use the Site Overview › Issues tab and the Technical Audit Test References to read each check in detail.

Page Overview tab

The default view when you open the report. Walks you through the page’s health in roughly the same order as you’d read it top-to-bottom on the screen.

1. Poli AI Insight

A blue banner at the top of the report. Poli AI reads the live page data and writes a short, page-specific summary - the most important finding and what to do about it. When Poli AI isn’t connected yet, the banner reads “Connect Poli AI to get a personalised insight for this page.”

The banner mirrors the same Poli AI Insight pattern you already saw on the GEO Audit Overview and the Site Overview tabs.

2. Indexability

The first of three summary cards at the top of the report. Tells you whether Google can find and index this page.

• Score out of 100 - e.g. 81/100, shown as a circular ring.
• Indexable: YES / NO - a small badge that answers the question directly.
• Status badge - Excellent, Good, Average, or Weak.

3. Page Status

The second summary card. Tells you whether the page actually loads.

• HTTP status code - what the server returned when POLIRIS asked for the page:
  • 200 - the page loaded successfully.
  • 3xx - the page redirects to another URL.
  • 4xx - the page is blocked (403) or missing (404).
  • 5xx - the server returned an error.
• Accessible: YES / NO - can the page be visited without issues?
• Status badge - mirrors the Indexability badge (Excellent / Good / Average / Weak).

4. Links Overview

The third summary card. Tells you how well-connected this page is inside your site.

• Incoming / outgoing count - shown as a ratio (e.g. 256 / 394 links).
  • Incoming links - other pages on your site linking to this page. More incoming links = more important to your site.
  • Outgoing links - links from this page out to other pages.
• Status badge - Excellent, Good, Medium, or Weak. A page well-connected on both sides scores higher.

5. Page metadata strip

Below the three summary cards sits a strip of compact page-level facts:

• CanonicalConflict
• RobotsAllowed
• SchemaOk
• Last crawlabout 1 month ago
• Incoming256
• Outgoing394
• Broken0

• Canonical - whether the page declares itself as the “official” version. OK, Missing, or Conflict.
• Robots - whether crawlers are permitted on this page. Allowed or Blocked.
• Schema - whether the page has structured data. OK, Partial, or Missing.
• Last crawl - how recently this page was last crawled. Green within 7 days, orange 8–30 days, red over 30 days.
• Incoming - number of internal pages linking to this one.
• Outgoing - number of links from this page to other pages.
• Broken - number of broken links found on this page.

6. Site Health Pipeline

The same pipeline pattern from Site Overview, here showing scores for this specific page. POLIRIS groups every page-level test into three categories and averages the results into a stage score.

• Page Access Can AI reach this page? Reachability, status codes, redirects, robots rules.
• Content Access Can AI read this page? Metadata, schema, headings, alt text, hreflang.
• Content Quality Is the content well-structured? Word counts, readability, uniqueness, depth.

Each stage shows an averaged percentage (e.g. Page Access 84%, Content Access 82%, Content Quality 53%). A Bottleneck chip appears above the weakest stage so you immediately know what’s holding the page back. In the example, Content Quality is the bottleneck - even though the other stages are healthy, the page can’t score better than its weakest link.

7. Overall Health Score

The overall combined score for this page sits in the top-right of the pipeline (e.g. 73%). It blends the three stage scores into a single number out of 100, so you have one figure to track per page over time.

8. AI Crawlers Access

A row of AI crawler chips showing which AI systems are allowed to crawl this page. Each chip carries a green dot when the crawler is permitted and a red dot when it’s blocked.

9. Top Issues

Below the pipeline, three columns list the most impactful issues found on this page - one column per Site Health stage. Each row shows the issue title, how many pages are affected (often 1 page at the page level), and a severity badge.

The severity colours mirror what you see across the rest of the audit:

• Critical Red - the most urgent issues. Fix first.
• Warning Orange - important issues to plan a fix for soon.
• Notice Blue - minor recommendations and polish.

10. Health Score Evolution

A chart showing how this specific page has changed over time. Use the toggle at the top to switch views:

• Health Score Page health over time as a percentage line. Trends up when fixes land.
• Issues Number of issues identified on this page over time. Trends down as you close issues.

Hover any point to see the exact value. If history isn’t yet available, the chart shows No history available yet.

11. Issues Overview

A small card on the right of the evolution chart summarising the total issue load for this page, broken down by severity:

• Total issues - the headline number (e.g. 24 issues).
• Critical - the urgent ones, fix first.
• Warning - important but not blocking, plan soon.
• Notice - minor and polish-level items.

Click Details on the card to jump into the per-issue breakdown. Working the critical row first is almost always the fastest way to lift the page’s overall score.

12. Page Speed Insights

A compact card showing the page’s vitals: Overall score plus the four Core Web Vitals POLIRIS measures - FCP, LCP, INP, CLS, and TTFB. Each metric is shown as a circular ring; higher / lower is better depending on the metric (e.g. LCP under 2.5s is good, CLS under 0.1 is good).

When no speed data is available yet for this URL, the rings render as zeros - usually a sign the page hasn’t been measured by PageSpeed yet rather than that it’s performing badly.

13. Robots.txt & Sitemap Coming soon

Two cards reserved for the per-page Robots.txt coverage and Sitemap inclusion check. The feature is in development - until it ships, both cards display Data not yet available with an Ask Poli AI shortcut so you can still ask questions about robots rules and sitemap coverage at the site level.

When the feature lands, each card will show the page’s status against your site’s robots rules and its presence in the sitemap, with one-click fix recommendations.

14. Connected Services

Three cards at the bottom of the page report unlock additional analytics when you connect external Google tools. Until connected, each card shows a blurred preview with a Connect button in the centre.

• Clicks vs. Impressions - how often this page appears in Google search results (impressions) and how many times it’s clicked, shown as a chart over time. Requires Google Search Console.
• Engagement - how visitors behave once they arrive: average engagement time, sessions, users, and engagement rate. Requires Google Analytics.
• Top Queries - the search terms bringing people to this page - up to four of the top queries by impressions. Requires Google Search Console.

The connection status uses the same dot logic used everywhere else in POLIRIS:

• Green dot - connected. The page-level analytics load with live data.
• Red dot - not connected. The audit still runs, but the connected-services cards stay blurred until you link the tool. See Connect Google Analytics & Search Console for the step-by-step guide.

Technical Audit Test References

A reference library of every check POLIRIS runs during a Technical Audit. You’ve already seen these tests summarised in Site Overview and Page Explorer - this page goes one level deeper for the curious. Each card is one test: what it looks for, what happens if it fails, and how to fix it.

• 102Total tests
• 96Page-level
• 6Site-level
• 15Critical severity

How to read each test

Every test is tagged with its level (whether it runs against your whole site or one specific page), a severity (how serious a failure is), a priority (how urgent the fix is), and whether the test is especially relevant for AI search. Together these tell you which fixes to tackle first.

Browse by Top Category

The tests are organised under the same Site Health Pipeline stages used in the audit dashboards. Jump to a category, or scroll through them all.

• Content Access54 tests across 11 subcategories
• Content QualityComing soon
• Page Access48 tests across 8 subcategories

Content Access

Tests in this group check whether the content on your pages is well-structured and readable for both search engines and AI tools - headings, metadata, schema, alt text, hreflang, and so on.

Content Quality Coming soon

The Content Quality stage of the Site Health Pipeline measures whether the words on each page are substantial, readable, and clear enough to be useful to a real person - not just whether the markup is technically correct.

Page Access

Tests in this group check whether your pages can actually be reached and crawled - canonicals, redirects, status codes, links, page speed, and pagination.

Putting It All Together

The Technical Audit and the GEO Audit (Visibility + Sentiment) are most valuable when you read them side by side. Each one explains the other - technical issues cap your GEO ceiling, and GEO results tell you which technical fixes to prioritize first.

Why cross-data wins

A great GEO score with a broken site is fragile. A perfect Technical score with no AI visibility is invisible. POLIRIS is built on the assumption that the two views belong in the same conversation.

A 4-step weekly workflow

The fastest way to turn POLIRIS into a habit is to run the same loop every week. It takes 30–45 minutes once your audits are scheduled.

1. Read the headlines

   Open Product Overview. Note the change in Visibility, Sentiment, and overall GEO score versus last week. This is your 60-second pulse.

2. Find the priority Category

   In Visibility, look at the Priority tag in Score Breakdown. In Sentiment, check the top High recommendation. If both surface the same Category, that’s your week’s focus area.

3. Pull the matching technical evidence

   In Page Explorer, look up the URLs that target that Category. In Site Overview › Issues, sort by severity. The technical issues blocking that Category are now your engineering and content backlog - with a GEO reason attached to each ticket.

4. Hand off with evidence

   Use Prompt Explorer to grab full AI answers and cited sources. Combine them with the technical issue list and ship one focused brief to your content, PR, and dev teams. Re-run the audits next week and watch the Priority Category move.

Pattern playbook

The combinations below show up over and over. Use them as a quick diagnosis when you don’t know where to start.

What to do next, by role

Each team should live in a different part of POLIRIS. Use this map to point people at the screen that answers their questions.

Implementation

Implementation is where audit findings turn into actual changes on your site. POLIRIS does the work for the SEO fixes it can apply directly via the tracking snippet, and writes the suggested change for everything else so you can paste it into your site or CMS.

What you find inside

• Two tabs at the top: Suggested fixes (one click) and Other audit issues (read only).
• A left sidebar that groups problems by category (titles, meta, headings, schema, performance, canonical, content quality…).
• A list of items in the middle, the full proposed change on the right.

Suggested fixes — one click, no code

Anything POLIRIS can fix directly in the page — page titles, meta descriptions, headings, Open Graph tags, JSON-LD schema, alt text — lives here. POLIRIS writes the new value for each affected page. You hit Apply and the change shows up on your site at the next page load. Dismiss if it’s not what you wanted.

Good to know

• The tracking snippet must be installed on the site for Apply to actually take effect. Without it, you can still browse suggestions, but they won’t go live.
• Nothing reaches the live site until you click Apply — the suggestion list is a safe staging area.
• Apply doesn’t touch your code. The change is layered on by the snippet when a visitor loads the page.
• Rollback removes a live change cleanly.

Other audit issues — read, copy, paste

The technical audit also flags problems the tracking snippet cannot fix on its own: page speed, sitemaps, internal linking, content quality, canonical tags, and more. You’ll find them here, one row per page affected, with the recommendation written in plain English.

For each issue you get

• The severity (Critical, Warning, Notice).
• The affected page URL.
• What the test checks and why it matters.
• The recommended fix.
• For content-related issues (passive voice, content depth, headings…), a Generate rewrite button that drafts the new version so you can Copy it into your CMS.

A typical run-through

1. Open Implementation from the left nav (the rocket icon).
2. On the Suggested fixes tab, click Generate suggestions. POLIRIS reads your latest audit and stages everything it can fix in one click.
3. Pick a category in the sidebar (Title tags, Meta descriptions, Heading optimizations, Structured data, Image alt text…). Click a row to read the proposed change on the right.
4. Hit Apply for the ones you’re happy with, Dismiss for the rest.
5. Switch to the Other audit issues tab for everything the snippet can’t fix on its own. For each one you care about, read the recommendation, hit Generate rewrite if the button is there, then Copy the result into your site or CMS.

Poli Agent

Poli Agent is the chat assistant built into POLIRIS. Ask it any question about your data in plain English (or French) and it answers using your own audit results - never invented numbers. Think of it as a small team of specialists you can talk to from one chat window.

What Poli Agent is

Poli Agent is one chat agent that speaks with five different specialist voices. Each voice has its own focus area, but they share the same chat window and the same data - you never have to pick which one to talk to. The agent picks the right voice for your question and tags its reply with that voice’s name and colour.

What it does

• Answers questions about your dashboards in plain language (e.g. “Why did our visibility drop in Germany last week?”).
• Summarises long reports into a couple of clear sentences.
• Suggests next actions tied to a real number on a real dashboard.
• Compares your brand to competitors across visibility, sentiment, and technical health.
• Hands off to Content Generation when you’re ready to actually write something.

How it works

You ask a question. Poli Agent reads the relevant parts of your POLIRIS data, reasons over them, and answers with prose, inline charts, and (when useful) one-click action cards - small buttons that let you follow up, open a deeper view, or approve a fix.

Every answer is grounded in your data. If a number isn’t in your audit, the agent says so - it never makes one up.

Four speed settings

A toggle in the chat input lets you pick how much thinking the agent does for each question:

• Auto POLIRIS picks the right speed for you. A good default.
• Fast Quick lookups and short follow-ups. Fastest, lowest cost.
• Balanced Multi-source questions, like comparing visibility and sentiment for a product.
• Deep Strategic, multi-step questions like “Build me a 90-day GEO plan based on this audit.”

Meet the team - the five voices

You always type in one chat window. The voice that replies changes depending on what you asked. Each voice shows its name, role, and a colour at the top of the message.

What Poli Agent knows

Poli Agent can securely access every dataset in your POLIRIS workspace. Your data stays private — the agent only sees what your account is allowed to see, and your information is never shared with other users or other workspaces.

• GEO visibility - how visible your brand is across each AI engine, per product, per zone.
• Sentiment data - Net Sentiment Score (NSS) and radar sentiment per axis and per competitor.
• Technical audit results - issues, scores, Core Web Vitals, robots, sitemaps, schema.
• Content analysis - semantic relevance, writing quality, AI readability for every page.
• Your brand structure - products, sites, social profiles, audiences, content preferences.
• Your site’s content map — the agent understands what every page on your website is about, spots pages that overlap or compete with each other, and surfaces content gaps or optimisation opportunities.
• The live web - through web search and static URL fetches when you ask about something external.

Optional connectors add even more: Google Search Console for impressions, clicks, CTR, and position; Google Maps for local-search and trade-area questions.

What you can ask

Examples to get you started. Mix product names, zones, engines, and time windows for sharper answers.

How to ask effectively

• Be specific Name the product, zone, engine, or time window. “Visibility in Germany on Perplexity last week” beats “how are we doing?”
• Ask why, not just what Dashboards already show the what. Poli Agent shines on “why did this drop?” and “which page is dragging us down?”
• Promote the speed for strategic questions Use Balanced or Deep for plans and multi-source synthesis - not for simple lookups.
• Click the action cards When a reply ends with a button, click it - the follow-up is pre-written for you.
• Edit to fork If a conversation drifts, edit your last message instead of starting over. The agent re-runs from that point.
• Use memory Tell Poli Agent something once and it remembers across sessions: “Remember: always answer me in French”, “We already cover X on /blog/x. Skip it next time.”

What it doesn’t do

So expectations stay realistic:

• It does not browse like a human - no clicks, no logins, no JavaScript rendering.
• It does not write full articles inline - Kate hands you off to Content Generation for full drafts.
• It does not edit your website directly. The only changes it can apply are small fixes delivered through a lightweight JavaScript snippet (the POLIRIS “pixel”) that you configure once inside My Site & Socials. Even then, every fix is shown to you as a card you have to click to approve before anything ships.
• It does not launch new audits. Visibility, sentiment, and technical audits are scheduled from the dashboard. Poli Agent reads results; it doesn’t kick off audits.
• It does not see across sites. Every session is scoped to one project.
• It does not invent data. If a number isn’t in your audit, the agent says so.

Content Generation

Content Generation is POLIRIS’s writing room. Talk to Kate, our content assistant, in plain language - she turns “I need to write something” into a finished, scored, publish-ready article in about a minute and a half.

Meet Kate

Kate is the AI assistant that runs the writing flow. She already knows your brand - your products, your tone, the pages you’ve already published, the topics your competitors cover, and the keywords people are actually searching for. You don’t fill in a form; you have a short conversation, and she does the work behind the scenes.

What it does

You start with something vague (“generate me a content”) and Kate walks you through five quick steps to a finished article:

1. Suggest topics

   Kate looks up your brand data and offers three topic ideas, each tied to a real gap or strength. You don’t need to know keywords yet - just pick the one that sounds right, or click “Give me more ideas.”

2. Pick audience and format

   Two short questions: who’s it for? and what kind of article? (guide, comparison, how-to, listicle). Each question has a “What do you recommend?” option - Kate checks what’s already ranking on Google and picks the best fit.

3. Choose an angle

   Kate writes three different article angles on the same topic (e.g., a styling guide, a brand-history piece, a care manual) and you pick one. Different angles = different audiences and different keyword targets.

4. Confirm the brief

   Kate pre-fills everything - title, audience, tone, length, primary keyword, secondary keywords. You just review and click Continue. Add real product facts in the Grounding Data box and Kate is allowed to cite them; leave it empty and she uses careful, non-fabricated language.

5. Get the article

   About 90 seconds later, you have a fully-written article with proper headings, internal links, and a quality score (typically 88–92 out of 100 on the first pass).

How it helps

• No blank page You never start from zero. Kate proposes options, you choose.
• Brand-aware Kate reads your existing pages and content gaps before suggesting topics, so she never proposes something you’ve already covered.
• SEO baked in Primary and secondary keywords, internal links, headings, and meta tags are filled in for you - aligned with what’s actually ranking on Google.
• No invented stats Kate won’t make up numbers. If a fact isn’t in the Grounding Data box, the article uses careful language (“research consistently shows…”) instead of inventing a percentage.
• Scored before publishing Every article is graded across 5 quality categories. If the score is low, Kate automatically rewrites the weakest sections and re-scores.

What you can create

• Guides - long-form articles that explain a topic in depth.
• Comparisons - “best X for Y” style pieces with tables.
• How-tos - step-by-step instructional content.
• Listicles - numbered list articles for quick reading.
• Product pages - description copy aligned to your brand voice.
• FAQ sections - answer blocks AI can quote directly.

The Content Score

Every article is graded out of 100 across five categories so you know it’s ready to publish:

• Keywords & Links Did the primary keyword land in the right places (intro, headings, conclusion)? Are there links out to authority sites?
• Topic Depth Did the article cover the same subtopics top-ranking pages cover?
• Readability Sentence length, passive voice, complexity - is it easy to read?
• AI Engine Ready Is the article optimised for AI search engines? Clear definitions, dated claims, comparison tables.
• Writing Quality Does it sound human and engaging? The only category that uses an AI judge - the rest are deterministic checks, so the same article gets the same score every time.

Articles scoring 90+ are flagged Ready to publish. If a score comes in lower, Kate automatically reviews the gaps and rewrites the weakest sections before showing you the result.

How to use it effectively

• Start vague, let Kate guide you. “Generate me a content” is a perfectly good first message - she’ll suggest topics tailored to your brand.
• Use “What do you recommend?” when you’re unsure about audience, tone, or format. Kate picks based on what’s actually ranking on Google for your topic.
• Paste real facts into Grounding Data. Product specs, prices, certifications, expert quotes - anything Kate can cite directly. The more grounded data, the more concrete the article.
• Read the score breakdown. If something dropped points, the score detail tells you exactly which terms were missing or over-used - useful when you’re editing the draft.
• Always review before publishing. Kate is fast and accurate, but you’re still the editor.

What you get

• A full draft you can edit inside POLIRIS.
• SEO meta tags (title, description, slug) generated automatically.
• Downloads as Word, Markdown, or HTML.
• One-click publish to WordPress, Webflow, or Shopify (if connected).

Glossary

Every word you’ll hear in POLIRIS, in one place - plus a broader glossary of GEO and AI terms used across the industry.

POLIRIS Terms

Product-specific terms used across the POLIRIS platform.

Product

Something your brand sells. The thing POLIRIS looks for in AI answers.

Trading Zone

A country or region where you sell. AI answers can change from zone to zone.

Request

A question you want POLIRIS to ask the AI. Usually written like a real customer would type it.

LLM

“Large Language Model.” The AI brains - ChatGPT, Claude, Gemini, Perplexity, Mistral, and more.

GEO

Generative Engine Optimization. Making sure AI tools mention and understand your brand. See the longer industry definition in the GEO & AI Glossary below.

Visibility Audit

Part of the GEO Audit. Counts how often AI mentions your brand compared to competitors.

Sentiment Audit

Part of the GEO Audit. Looks at whether AI talks about you positively, neutrally, or negatively.

Poli Agent

The POLIRIS AI assistant. Helps you navigate the platform, run audits, and summarize results.

Audit

A full run of your Requests across your Zones and LLMs, with a report at the end.

Dashboard

The main screen with charts and numbers. One audit = one dashboard.

Plan

Your subscription: Starter, Growth, or Scale. Decides your monthly limits.

Organization

Your top-level POLIRIS account. Holds your branding, members, billing, and clients.

Workspace

One project inside your Organization - usually one brand or one client.

Client

A brand you work with (agency mode). Each Client has its own Workspace.

Prospect Client

A brand you don’t work with yet. Run a light audit to pitch them.

Member

A person on your team. Roles: Owner, Admin, Editor, Viewer.

Request Prompt

The behind-the-scenes template POLIRIS uses to talk to an LLM. A Request is what you ask; the Request Prompt is how POLIRIS asks it. For the general AI definition of “prompt,” see the GEO & AI Glossary below.

Visibility Area

The combination of Trading Zones, audiences, and LLMs where POLIRIS watches for mentions of your brand.

GEO & AI Glossary

A broader glossary of Generative Engine Optimization and AI terminology used across the industry. Useful when you’re reading reports, talking to partners, or comparing POLIRIS output to other tools.

AI Answer Box / AI Summary / AI Snapshot

AI-generated boxes on results pages that provide a summary or direct answer to a query. They often combine information from multiple sources, similar to AI Overviews, and also include links to reference content. See also: AI Overview, Answer Engine, Enhanced Results.

AI Overview

AI-generated summaries displayed on Google results pages, combining information from multiple sources. They provide a quick overview of a topic, with links to relevant content, often in a prominent area at the top of the SERP. AI Overviews extend the Search Generative Experience and rely on Google’s advanced models. See also: SGE, AI Mode, AI Answer Box.

Answer Engine / Response Engine

A system that directly answers questions by computing or generating a response from external data (e.g., WolframAlpha). Unlike traditional search engines, it provides a precise answer rather than a list of links. It relies on knowledge bases, algorithms, and sometimes language models. See also: AEO, AI Answer Box, Answer Box.

Answer Engine Optimization (AEO)

Content optimization to appear in direct answers provided by answer engines or conversational assistants. AEO focuses on content structure, clarity, and reliability so that AI systems can accurately retrieve and cite it. It specifically targets answers generated by tools like ChatGPT or Google’s AI Overviews. See also: GEO, Answer Engine, Enhanced Results.

Chunking

The act of breaking large volumes of text into smaller units, called “chunks,” to make them easier for language models to process. This improves the efficiency and accuracy of summarization, information extraction, or translation. Chunks may consist of sentences, paragraphs, or passages, in order to preserve relevant context. See also: Passage, Context Window.

Citation

Explicit mention of a source or document in the response generated by a model, used to reference the origin of the information. RAG systems and AI Overviews use citations to show which pages were consulted and to strengthen user trust. Structuring your content and providing reliable sources increases the chances of being cited. See also: RAG, AI Overview, E-E-A-T.

Context Window

The maximum number of tokens a model can consider simultaneously around a word or query. A larger context window allows the model to retain more information and can improve response relevance. It determines how much text an LLM can analyze at once and influences chunking and RAG strategies. See also: Token, LLM, Chunking.

Conversational Search

Search where the user asks questions in natural language and receives responses as in a conversation. It replaces keyword-based searches with full queries and offers a more human-like interaction. These systems rely on language models and context to understand intent and provide evolving answers (often via chatbots or voice assistants). See also: Answer Engine, Search Intent, Prompt.

Data ingestion

The process by which raw data is collected, imported, and integrated into a system for processing or analysis. For language models, this includes gathering texts, cleaning them, and formatting them for training or updating via techniques like RAG. Effective ingestion ensures the quality and freshness of the information used by generative engines. See also: RAG, Grounding, Vector Databases.

E-E-A-T

Acronym for Experience, Expertise, Authority, Trustworthiness. This is a content quality evaluation framework used by Google. Experience reflects first-hand knowledge of the topic; expertise refers to skills or qualifications; authority concerns reputation and the quality of sources; trustworthiness relates to site security and transparency. Strengthening these criteria improves content credibility and visibility, even though they are not direct ranking factors. See also: YMYL, LLMO, Rich results.

Embeddings

Vector representations (in numerical form) of words, phrases, or objects, organized so that semantically similar items are close in the vector space. Obtained through learning techniques, they allow measurement of semantic similarity between texts or entities. They are used for semantic search, sentiment analysis, and information retrieval in vector databases. See also: Semantic similarity, Vector database, Semantic search.

Enriched Results

Search results that go beyond the traditional blue link, displaying additional data, images, or visuals (reviews, products, events, etc.). They are often generated from structured data. They enhance the user experience and can achieve high click-through rates, even with the rise of zero-click searches. See also: Structured Data, Rich Snippet, Featured Snippet.

Entity / Named Entity

An important element of a text (person, place, organization, event, date, etc.) that can be detected and categorized using named entity recognition techniques. Identifying entities feeds knowledge graphs and improves search relevance. In GEO, properly defining and tagging entities enhances their recognition by models and increases the likelihood of being cited. See also: Knowledge Graph, Schema.org, Embeddings.

Featured Snippet / Position Zero

A highlighted text snippet at the top of a search results page that provides a concise answer to a query. It is called “position zero” because it appears before the organic results, often in a box. It is selected from well-structured and relevant content, making it important for SEO and GEO. See also: Enhanced Results, Zero-Click Search, AI Answer Box.

Generative AI

A set of models capable of creating text, images, or other original content from existing data. Often based on transformer architectures and LLMs, generative AI produces responses, summaries, or multimodal creations and powers conversational assistants and creative tools. It also raises reliability and ethical concerns. See also: LLM, Foundation AI, Hallucination.

Generative AI Optimization (GAIO)

Term used to describe the optimization of content to adapt it to generative engines (the conceptual equivalent of GEO). It combines structuring, markup, and semantic relevance so that language models recommend and cite a brand’s content. GAIO largely overlaps with GEO. See also: GEO, LLMO, AEO.

Generative Engine Advertising (GEA)

Adaptation of advertising strategies to optimize brand visibility and recommendations within AI-generated responses. This involves creating promotional content that is likely to be selected and mentioned by generative engines such as ChatGPT or Perplexity. GEA extends GEO by specifically targeting the commercial recommendations made by the models. See also: GEO, LLMO.

Generative Engine Optimization (GEO)

The optimization of digital content to improve its visibility in results generated by AI models, particularly the synthetic answers produced by generative engines. This practice aims to influence the way large language models retrieve, synthesize, and cite a brand’s or publisher’s information in generated responses. It resembles an evolution of search optimization toward environments where users receive a complete answer without clicking, and it is distinct from Search Engine Optimization (SEO) and Answer Engine Optimization (AEO). Also called: GAIO, AI SEO, generative engine optimization. See also: AEO, RAG.

Generative Search Optimization (GSO)

A set of optimization techniques aimed at ensuring content is effectively recognized by search engines using generative AI. Similar to GEO, AEO, and GAIO, it emphasizes semantic quality and the accessibility of information for the models, whether in generated answers, enriched results, or knowledge panels. See also: GEO, Global GSO, LLMO.

Global Search Optimization (GSO)

A set of optimization practices aimed at improving the visibility of content across all search interfaces: traditional search engines, voice assistants, generative engines, etc. It combines SEO, AEO, and GEO to ensure presence in organic results, direct answers, and generated previews. See also: SEO, GEO, LLMO.

Google AI Mode

An advanced search experience using deep reasoning and multimodal capabilities to explore a topic in detail. It allows follow-up questions, provides links to useful resources, and integrates AI-generated answers. AI Mode notably relies on the query fan-out technique to break a query into sub-questions and explore the web in depth. See also: Query Fan-Out, SGE, AI Overview.

Grounding / Factual Anchoring

The act of linking an AI model’s output to real, verifiable data to ensure factual results. Grounding improves accuracy by basing generation on authentic documents rather than mere statistical correlations. It is often used alongside RAG to combat hallucinations and provide clear citations. See also: RAG, Hallucination, Generative AI.

Hallucination (AI)

Inaccuracy or false statement generated by an AI model, while appearing plausible. It occurs when the model misinterprets patterns or lacks reliable information, producing responses that do not reflect reality. Reducing hallucinations involves techniques such as RAG or grounding, which link the output to verifiable sources. See also: Grounding, RAG, Generative AI.

JSON-LD

Serialization format for linked data that allows structured data to be embedded within HTML using JSON. Recommended by Google for implementing Schema.org without modifying visible content, it simplifies markup maintenance and promotes the appearance of enriched results and knowledge panels. See also: Structured Data, Schema.org, Knowledge Graph.

Knowledge Graph

A knowledge base that represents entities and their relationships as nodes and edges. It stores interconnected descriptions of objects, events, and concepts, allowing navigation between them via queries. Search engines use it to power knowledge panels and generated answers. See also: Named Entity, Schema.org, Knowledge Panel.

Knowledge Panel

Box in search results that presents a summary of information from a knowledge graph (key facts, images, links, etc.) about an entity. Optimizing its presence involves using structured data and precise schemas for the relevant entities. See also: Knowledge Graph, Schema.org, Enriched Results.

Large Language Model (LLM)

Large language model trained in a self-supervised manner on very large volumes of text for natural language processing tasks. Based on transformer-type architectures, it contains billions to trillions of parameters and can generate, summarize, translate, and reason over text. LLMs are at the core of conversational agents, code generators, and augmented search systems. See also: Generative AI, Transformers, Context Window.

Large Language Model Optimization (LLMO)

Optimization of content so that it is used and cited by generative AI tools (ChatGPT, Perplexity, etc.). It prioritizes authoritative sources, semantically complete content blocks, and formats that are easy to analyze. LLMO is a variant of AEO and GEO focused on large language models. Also called: AI SEO, LLM SEO. See also: GEO, AEO.

Multimodal Model / Multimodal Search

AI model capable of understanding and processing multiple types of information (text, image, audio, video) simultaneously. Multimodal search leverages these models to interpret queries combining multiple media and provide integrated results (e.g., describing an image and asking a question). See also: Foundation AI, LLM, Transformers.

Natural Language Processing (NLP)

A field that enables machines to understand, analyze, and generate human language. It covers syntax analysis, translation, entity recognition, sentiment analysis, and more. Recent advances rely on LLMs, embeddings, and pretraining to enhance human–machine interaction. See also: LLM, Transformers, Embeddings.

Ontology

Formal representation of the concepts within a domain and the relationships that link them, used to structure knowledge. It defines classes, properties, and hierarchies to ensure a shared understanding (for both machines and humans) of a subject. Ontologies are crucial for knowledge graphs and support semantic search and enriched results. See also: Knowledge Graph, Taxonomy, Schema.org.

Passage

Text segment short enough to be processed or indexed individually. Passages are often created through chunking and serve as units for search or input into generative systems. A well-defined passage improves accuracy by focusing on a specific context. See also: Chunking, RAG.

Prompt

Natural language text that describes the task an AI model should perform. It provides the initial context and can specify the style, format, or information to use. The quality and precision of the prompt directly influence the relevance of the responses and form the core of prompt engineering. See also: Prompt Engineering, Query Fan-Out, Conversational Search.

Prompt Engineering

The art of formulating and structuring an instruction (prompt) to obtain more relevant responses from a generative AI model. A prompt describes the task to perform, sometimes including the desired style, format, or context. Optimizing it is essential to fully leverage language models in both search and creative applications. See also: Prompt, Query Fan-Out, Conversational Search.

Query Fan-Out

Technique that breaks a question into multiple subtopics and simultaneously sends several queries to find diverse content. It helps capture different intents and retrieve results from the web, knowledge graphs, and other sources. Query fan-out powers generative search experiences such as AI Mode. See also: AI Mode, SGE, Conversational Search.

Retrieval-Augmented Generation (RAG)

Technique that combines a language model with an information retrieval system to incorporate up-to-date knowledge into generation. The model first consults a set of specific documents before responding, which enriches the output and reduces hallucinations. It allows updating a model without full retraining and facilitates verifiable citations. See also: GEO, Grounding, Augmented Retrieval.

Schema.org

Collaborative initiative aimed at defining schemas for structured data on the Internet. Led by players such as Google, Microsoft, Yahoo, Yandex, and the web community, it provides common tags to improve search engines’ understanding of pages. Using Schema.org in HTML helps generate enriched results, knowledge panels, and accurate citations. See also: Structured Data, JSON-LD, Knowledge Graph.

Search Generative Experience (SGE)

Experience from Google Search Labs that uses generative AI to provide quick overviews of a topic, related ideas, and the ability to ask follow-up questions. SGE presents synthesized answers accompanied by links to sources, helping users understand a topic more quickly. It paves the way for more conversational search and has inspired AI Overviews and AI Mode. See also: AI Overview, AI Mode, Query Fan-Out.

Search Intent

What the user actually wants to achieve when entering a query: informational (seeking a fact), transactional (purchasing or booking), or navigational (accessing a specific site). Understanding the intent allows content and responses to be tailored to better meet the need. See also: Conversational Search, Query Fan-Out, Enriched Results.

Search or Crawl Bots

Programs that automatically browse websites to download and index their content. They allow search engines to discover and update pages, but also serve to collect data for training language models. Publishers can control their activity using the robots.txt file and certain meta tags. See also: Knowledge Graph, LLMO, Web Crawling.

Semantic Proximity

Measure of the similarity in meaning between two words, phrases, or entities within a vector space. Calculated from embeddings, it allows finding relevant content beyond simple keyword matching. It is essential for semantic search and content alignment in answer engines. See also: Embeddings, Semantic Search.

Semantic Search

Information retrieval approach that aims to understand intent and contextual meaning rather than relying solely on exact keywords. It leverages embeddings and measures semantic similarity to deliver more relevant results, particularly for conversational queries. See also: Embeddings, Semantic Proximity, Vector Database.

SEO LLM / LLM SEO

Informal term referring to the optimization of content so that it can be utilized and cited by large language models. Similar to LLMO, it involves structuring content, providing reliable sources, and maximizing the models’ semantic understanding. It combines traditional SEO and GEO to ensure visibility in generated answers. See also: LLMO, GEO, AEO.

Structured Data

A standardized format used to provide information about a page and classify its content (e.g., a recipe: ingredients, cooking time, nutritional value). This markup helps engines understand the content and supports the generation of enriched results and the use of the page by language models. See also: Schema.org, JSON-LD, Enhanced Results.

Taxonomy

Hierarchical classification used to organize concepts, products, or content into categories and subcategories. It facilitates user navigation and search engine indexing. In GEO, a consistent taxonomy helps models understand relationships between topics and generate more accurate responses. See also: Ontology, Knowledge Graph, Structured Data.

Token

A sequence of characters considered a processing unit during tokenization (word, subword, symbol, etc.). Tokenization splits text into tokens, which the model then processes. The number of tokens affects the context window and processing costs. See also: Context Window, Transformers, Embeddings.

Transformers

Neural network architecture based on the multi-head attention mechanism, without recurrence. Text is tokenized and embedded, then each token is contextualized via attention weights within a context window. Introduced in the paper “Attention Is All You Need” (2017), transformers have replaced RNNs and form the basis of most large language models. See also: LLM, Context Window, Token.

Vector Database

A database designed to store and query embeddings (vectors representing the meaning or features of unstructured data). Unlike traditional databases, it handles multidimensional data and searches by similarity (distance between vectors). It powers semantic search and RAG by providing relevant documents through vector queries. See also: Embeddings, Semantic Search, RAG.

YMYL (Your Money or Your Life)

Category of content that can impact users’ health, safety, happiness, or financial stability. Google applies higher quality standards to these pages to prevent inaccurate information from harming users. Publishers must demonstrate strong expertise and trustworthiness to be visible in these sensitive areas. See also: E-E-A-T, Enriched Results, LLMO.

Zero-Click Search

Search where the answer to the query is provided directly on the results page, without the need for an additional click. This can take the form of rich snippets, knowledge panels, or generated answers that immediately satisfy the search intent. This trend impacts GEO and AEO, as it reduces site traffic in favor of presence in the immediate answer. See also: Enriched Results, AI Overview, Featured Snippet.

FAQs

Generative Engine Optimization is still new for many of you. Below are answers to the most frequently asked questions about our methods, deliverables, and how POLIRIS supports you.

What is GEO?

Generative Engine Optimization (GEO) refers to all techniques aimed at optimizing a brand’s presence in responses generated by AI engines (LLMs). In practical terms, this involves optimizing your content, site structure, and reputation signals so that AIs can easily find you, cite you accurately, and represent you faithfully according to your positioning.

Which AI platforms are supported?

We analyze your visibility exclusively on the main AI engines (general-purpose LLMs): those that generate comprehensive answers using the web and their training data - ChatGPT, Perplexity, Gemini, Mistral, Claude, and Deepseek.

We do not audit specialized assistants or vertical AIs (for example, assistants integrated into highly specific business tools). The goal is to measure your presence where your customers ask “general” questions about your market, products, or competitors.

What does the scoping audit include?

The scoping audit is a quick initial overview of your presence in AI engines. It specifically includes:

• a visibility ranking of your brand vs. your main competitors,
• the top sentiment qualifiers used by AI to describe you,
• an accessibility status (to what extent your content is discoverable / usable by LLMs).

It is a preliminary assessment, designed to decide on action priorities and determine whether a full audit is appropriate.

What does the full audit include?

The full audit provides a comprehensive and actionable view:

• a multi-LLM analysis (overall and by segment / product / market),
• a detailed perception: brand radars, mapping of strengths / weaknesses, main qualifiers used by AI,
• an Access / Structure / Content audit of your key pages, with scores, prioritization, and a GEO action plan.

This is the basis for a genuine GEO roadmap for your brand.

What is the difference between a scoping audit and a full audit?

Scoping audit

• Quick, focused on overall visibility, sentiment, and accessibility.
• Ideal for understanding your position relative to competitors and setting priorities.

Comprehensive audit

• In-depth analysis across multiple LLMs and segments.
• Detailed diagnosis of Access / Structure / Content, brand perception, and a structured action plan.

In short: the scoping audit is used to decide where to go, and the comprehensive audit to decide exactly what to do.

How do you assess AI sentiment?

We analyze AI engine responses to extract:

• the key attributes associated with your brand,
• the strengths highlighted,
• the weaknesses or areas of uncertainty,
• any potential gaps with your current positioning.

We aggregate these indicators, then we contextualize them against your positioning and brand. The goal is twofold:

• to correct what is still circulating (outdated data, biased or incomplete perceptions),
• to emphasize your strengths (additional evidence, reference content, client cases, etc.).

How do you measure visibility?

We build a set of representative test queries for your market:

• brand queries,
• product / service queries,
• “problems / needs” queries from your customers.

We test them on the main LLMs. The results are then consolidated to produce:

• visibility rankings (your brand vs. your competitors),
• explicit gaps by product, by segment, and by AI engine.

You can see very quickly where you appear strongly, where you are absent, and where competitors dominate.

What action plans does POLIRIS propose after the audit?

After the audit, you don’t just receive scores - you receive a concrete action plan. At POLIRIS, we provide you with clear and easy-to-implement recommendations to boost your presence in the AI landscape:

• how to optimize your content so it is understood and cited by AI engines,
• how to select and structure your keywords and topics to align with user intents,
• which gaps and competitive opportunities to prioritize first.

What are the timelines?

For the scoping audit, you receive the results within 24 hours after the scope is validated and the necessary information is received.

For the comprehensive audit, the duration depends on the scope being analyzed (number of brands, markets, languages, competitors…). The steps and milestones - kick-off, data collection, interim analyses, reporting - are defined with you at the start of the project.

What deliverables do you receive?

At the end of a GEO audit, you receive:

• a detailed report (PDF and/or slides),
• detailed and prioritized recommendations, structured by pillar and by segment,
• a feedback workshop with your teams.

Optionally, we can also set up:

• a monthly dashboard to track GEO visibility,
• regular monitoring of the evolution of your position versus competitors.

Which markets and languages do you cover?

We operate worldwide. Our methodology adapts to the required language: we define with you the priority markets, the languages to audit, and the most relevant AI engines for each region.

Do you work with SEO agencies?

Yes, on a B2B2B basis.

We co-build the GEO action plans with your partners: SEO agencies, integrators, internal web teams…. The goal is to:

• translate GEO insights into concrete actions (technical, content, popularity),
• accelerate execution, relying on the teams already in place rather than replacing them.

How is the score calculated per pillar?

For each tested URL, we evaluate three dimensions:

1. Access - the ability of AI engines to find and use the page,
2. Structure - how the information is organized and tagged,
3. Content - the quality, clarity, and alignment with search intent.

Each page receives a score for Access / Structure / Content. These scores are then consolidated at the domain, product, or segment level to give you an overall view and priorities.

What ROI can be expected?

GEO primarily aims to improve your visibility and credibility in generated responses. It is therefore difficult to quantify immediate ROI in terms of sales, but you can expect:

• an increase in qualified traffic from AI and search engines that incorporate generative answers,
• a reduction in support costs thanks to more comprehensive automated responses,
• an improvement in your online reputation.

Results vary depending on your starting point and the speed of implementing the recommendations.

Why is GEO essential?

The web no longer informs only humans - it now also feeds AI. Traditional SEO tools, however, are not suited for AI, which is why a dedicated GEO approach is essential.

Is my site’s data secure?

Yes. We primarily work on your public content (website, product pages, editorial content) and, if needed, on documents that you choose to share with us. We do not access your internal systems (CRM, back-office, analytics tools…) without explicit consent, and the data is never sold or reused outside of your audit. All analyses are conducted within a framework compliant with GDPR, with access restricted to the project team.

Support

Stuck or have feedback? We’d love to hear from you.

• 📧 Email: support@poliris.io
• 💬 In-app chat: the bubble in the bottom-right corner.
• 📚 More guides: poliris.io/resources

Administrator Settings

The settings area is the control room of POLIRIS. From here, admins tune how the platform behaves for everyone on the team - from logos to LLMs, from who can log in to how much is spent.

Two levels of settings

POLIRIS is split into two layers. The same team can work across both.

Organization Settings

The “name tag” of your account. This is where you set the company’s name, website, industry, headquarters, and contact details.

POLIRIS uses this info to pre-fill audit defaults and to label exports and client reports.

Branding

Upload your logo, pick your primary color, and choose a default font. These settings are applied to shared dashboards, PDF exports, and client-facing reports.

What you can customize

• Logo (light + dark variants)
• Primary & accent colors
• Typography / heading style
• Favicon for shared dashboard links
• Custom domain for hosted reports (Scale plan)

Clients

If you’re an agency or consultancy, Clients is where every brand you work with lives. Each Client gets its own Workspace, its own Products, and its own reports.

What you can do here

• Create a new Client from a website URL.
• Archive an old Client without deleting its data.
• Invite the Client’s own team to view (read-only) their dashboard.
• Switch between Clients with one click.

Prospect Clients

Prospect Clients is the “pitch shelf.” It’s for brands you don’t work with yet - leads, demos, and sales pitches.

You can run a light audit on a Prospect to show them what their GEO score looks like today, then convert them into a real Client with one click when the deal closes.

Members

The Members panel is your team list. Here you invite coworkers, change their roles, or remove them when they leave.

Roles

Billing

Billing is where you pay - and get paid back if needed. You can update your card, switch plans, download invoices, and buy credit top-ups.

What lives here

• Current plan (Starter Growth Scale) and renewal date.
• Payment method - card, SEPA, or invoice.
• Billing address and VAT info.
• Invoice history (downloadable as PDF).
• Credit top-up packs.

Usage

Usage is the odometer of your POLIRIS account. It shows how many credits your team has spent this month, on what, and by whom.

What you can see

• Credits used vs. credits remaining, by month.
• Breakdown by action (LLM queries, technical scans, content drafts).
• Breakdown by Client or Workspace (if you have several).
• Trend charts so you can predict when you’ll run out.

Workspace Settings

A Workspace is one project inside POLIRIS - usually one brand or one client. This panel holds the settings that shape that one project.

What lives here

• Workspace name, description, and time zone.
• Default language and market.
• Default Trading Zones and LLMs for new audits.
• Data retention (how long audit history is kept).
• Archive or delete the Workspace.

Configuration

Configuration is the “engine room.” It’s where admins fine-tune how audits run behind the scenes - the same settings the onboarding wizard guessed for you on day one.

Common tweaks

• Default audience profile (B2C, B2B, agency…).
• Depth of each audit (quick / standard / deep).
• Retry behavior when an LLM times out.
• Competitor shortlist to always compare against.

Products (admin view)

This is the master list of every Product in your Workspace. The regular Product editor lets you add and edit one Product at a time; the admin view lets you manage them all at once.

What you can do here

• Bulk activate, deactivate, or archive Products.
• Reorder the default priority for audits.
• Export the full Product catalog as CSV.
• Import Products from CSV or an e-commerce feed.

LLM Models

The LLM Models panel is where you pick which AI brains POLIRIS may talk to - and how.

Per-model controls

• Turn a model on or off for this Workspace.
• Choose the model version (e.g., GPT-4o vs. GPT-4.1).
• Set temperature (how creative vs. predictable).
• Bring your own API key, or use POLIRIS’s bundled access.

Prompts

Prompts are the templates POLIRIS uses behind the scenes. A Request is what you ask; a Prompt is how POLIRIS asks it to an LLM.

What you can do here

• Edit the system prompt used for GEO queries.
• Tweak the prompt used to write content drafts.
• Add custom prompts for recurring tasks (e.g., “summarize this audit in French”).
• Reset a prompt back to the POLIRIS default.

Visibility Area

The Visibility Area is the “stage” where POLIRIS watches for mentions of your brand. It combines your Trading Zones, your chosen audiences, and your tracked LLMs into one clean definition.

Why have it as its own panel?

• One place to see every region, audience, and LLM you cover.
• Spot gaps - a market or model you haven’t added yet.
• Compare visibility side-by-side across areas.

Scheduling

Scheduling lets POLIRIS run audits for you on a calendar - daily, weekly, or monthly. Set it once, and fresh dashboards land in your inbox automatically.

What you can schedule

• Recurring GEO audits (per Workspace, per Product, or per Zone).
• Weekly technical scans.
• Monthly content-draft batches.
• Alert emails if visibility drops more than X%.

Integrations

Integrations connect POLIRIS to the tools your team already uses - so dashboards, alerts, and data can flow both ways.

What you can plug in

• Slack - alerts when visibility changes or audits finish.
• Google Search Console - combine classic SEO with GEO.
• Google Analytics - attribute traffic gains to POLIRIS actions.
• WordPress / Webflow / Shopify - publish content drafts in one click.
• Zapier / webhooks - pipe audit results to anything.
• API key - build your own dashboards.

Logs

Logs is the black-box recorder of your account. Every meaningful action - who ran an audit, who edited a Prompt, who invited a member - shows up here with a timestamp.

Useful for

• Debugging (“why did that audit not run yesterday?”).
• Compliance reviews.
• Tracing a change back to the person who made it.
• Investigating unusual credit spikes.

Release Notes

What’s new in POLIRIS. Fresh features, fixes, and small tweaks - in plain language.

April 2026

• New: Poli Agent now available for every plan.
• New: Sentiment Audit adds per-theme breakdowns.
• Improved: Workspace switcher is 3× faster.
• Fixed: Scheduled audits no longer skip February 29.

March 2026

• New: Administrator Settings hub - 17 panels in one place.
• New: Prospect Clients gets its own credit pool.
• Improved: GEO dashboard loads on mobile without crashing.

February 2026

• New: Shopify and Webflow integrations for one-click publishing.
• Improved: Content drafts support more languages (now 12).
• Fixed: Branding colors now export correctly in PDFs.