The essentials
- What we measure: for a list of questions, the presence of your site or brand in each engine's answers, its rank, the competitors and sources cited and the sentiment, with repeated checks.
- Manual: feasible for 10 to 20 questions a month with a spreadsheet; enough to get started.
- Tools: GEO specialists ($29 to $300 per month at entry level, enterprise on quotation), AI modules in SEO suites (Semrush, Ahrefs), management platforms that combine tracking with the other channels.
- Criteria: engines covered, query repetition, share of voice, France localisation, export, price per question.
AI engines have no Search Console: to find out whether ChatGPT, Perplexity or Gemini cite you, you have to ask them questions and measure their answers. This is at the heart of measuring GEO. This article describes what to measure, the manual method to get started, the families of tools available in 2026 with their prices, and the selection criteria. The metrics themselves are defined in GEO KPIs.
What to measure
| Data | Detail | Why |
|---|---|---|
| Presence | Your domain cited as a source, or your brand named, for each question and each engine | Citation rate and mention rate |
| Rank | Position of your source in the cited list | A first source carries more weight than an eighth |
| Competitors and sources cited | Who appears instead of you or alongside you | Share of voice, targets to emulate |
| Content of the mention | What the AI says about you: accurate, dated, favourable | Accuracy and sentiment |
| Variability | Results over several runs of the same question | Answers change; a reliable measurement averages several checks |
| Trend | The same questions over time | Impact of actions |
The manual method to get started
- Build the list: 10 to 20 target questions, split between understanding, comparing and choosing; see keyword research for GEO.
- Choose the engines: ChatGPT with web search enabled, Perplexity, Google (AI Overview and AI Mode), and possibly Gemini and Copilot.
- Check once a month, in a private window or with a neutral account (memory and history personalise answers), each question twice per engine.
- Record in a spreadsheet: date, engine, question, cited (yes/no), rank, brands and domains cited, comment on the content.
- Calculate the citation rate per engine and the list of the most cited competitors.
Allow two to three hours a month for 20 questions across three engines. Beyond that, or to track competitors closely, a tool becomes necessary.
Tool families in 2026
| Family | Examples | What they do | Indicative price excl. VAT / month | Who for |
|---|---|---|---|---|
| Specialist GEO tools | Peec AI, Profound, Otterly, Scrunch, Rankshift, AthenaHQ, Semrush AI Toolkit (module) | Query the engines at regular intervals on your questions, measure citations, mentions, share of voice, sentiment and sources; competitor tracking | $29 (Otterly, 15 questions) to $300 (Scrunch, 350 questions) at entry level; Qwairy €79 excl. VAT; enterprise on quotation | Brands and agencies for which AI recommendations matter |
| AI modules in SEO suites | Semrush (AI Toolkit), Ahrefs (Brand Radar), Similarweb | Brand visibility in AI answers, keywords triggering AI Overviews, comparison with competitors | Included or as an add-on (Semrush AI Toolkit around $99) | Teams already equipped with a suite |
| Rank tracking tools with AI detection | Monitorank, SE Ranking, AccuRanker | Presence of AI Overviews on your Google queries, citations in the overview, sometimes checks on assistants | €10 to €150 | SEO tracking that wants to see the impact of AI Overviews |
| Marketing management platforms | GreenRed (GEO module) | Citations and mentions per engine, share of voice, linked to AI traffic (GA4), SEO rankings and other channels, with alerts and actions | Included in the subscription | SME leaders and agencies who want an overview |
| In-house solutions | Scripts using the OpenAI, Perplexity and Gemini APIs | Automated checks on demand | A few euros of API costs + development | Technical teams |
The market is moving fast: tools appear and merge every quarter; check the engines covered and the prices when you choose. Measurements taken via APIs do not exactly reproduce the answers of the consumer interface (parameters, personalisation, web search), which serious tools document.
The selection criteria
- Engines covered: at minimum ChatGPT, Perplexity, Google (AI Overviews and AI Mode) and Gemini; Copilot, Claude and Mistral's Le Chat depending on your audience.
- Checking method: query repetition (several runs per question), frequency (daily to weekly), interface or API, France localisation and French language.
- Share of voice: competitor tracking on the same questions.
- Source analysis: which pages and domains are cited, to guide your actions; see the sources generative AI uses.
- Accuracy and sentiment: what the AI says about you.
- Integration: connection with GA4 (AI traffic), Search Console and SEO tracking, so you do not multiply dashboards.
- Price per question: tools bill by number of tracked questions; 30 to 50 questions are enough for an SME.
- Export and history: to keep your data if you change tools.
Interpreting the results
Three precautions. Answers vary: judge on averages and three-month trends, not on a single check. Engines cannot be compared with each other: a citation rate of 30% on Perplexity and 10% on ChatGPT is common, so compare each engine with itself over time. Finally, cross-check with traffic: sessions coming from assistants in GA4 confirm that citations generate visits; see identifying AI traffic in GA4. Monitoring what AIs say about the brand, beyond the target questions, is covered in monitoring your presence in generative AI.
How GreenRed helps
Rather than juggling several tools, GreenRed's GEO tracking brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.
Frequently asked questions
Is there a Search Console for ChatGPT?
No. No assistant provides a tracking tool for website publishers. Measurement relies on regularly querying the engines on a list of questions, either manually or with specialist tools that automate the checks and calculate the citation rate, share of voice and cited sources.
How much does an AI visibility tracking tool cost?
From $29 to $300 per month at entry level for specialist tools depending on the number of questions and engines, €190 for the Semrush plan that includes AI tracking, €179 extra for Ahrefs Brand Radar, and a few dozen euros for AI Overview detection in a rank tracking tool. Some marketing management platforms include it in their subscription.
Can you track your AI visibility without a tool?
Yes, for 10 to 20 questions: ask them each month to ChatGPT, Perplexity and Google in a private window, twice each, and record presence, rank and competitors cited in a spreadsheet. Allow two to three hours a month. Beyond that, a tool automates the checks and makes them more reliable.
Why do a tool's results differ from what I see in ChatGPT?
Tools often query the engines via APIs or neutral accounts, whereas your interface personalises answers based on your history, location and memory. Answers also vary from one run to the next. Serious tools repeat queries and document their method; compare trends rather than isolated checks.