Resources
How we measure your AI visibility
The RecomendIA score (0 to 100) combines 60% AI visibility, measured with real buying questions to each engine with web search, and 40% site GEO readiness, measured with 29 technical checks. Bands: low from 0 to 39, medium from 40 to 69, high from 70 to 100.
AI visibility (60%)
We ask each engine 5 recommendation questions without mentioning your brand and 1 direct question about it, in the language and location you choose. All with web search on, through each engine’s official API. Questions are phrased for your type of business: if you type “clothing”, we ask about clothing stores or brands, not garments.
In the free check you see the answers to 4 questions per engine; the other 2 are shown in full with the Pro plan, which also uses 6 recommendation questions.
Visibility measures only the recommendation questions, that is, what happens when a customer asks for options without naming you. If no engine recommends you, your visibility is 0. The direct question about your brand is reported separately, as “Does AI know who you are?”, and does not add to this score.
| Component | Points |
|---|---|
| Answers that recommend you | 50 |
| Rank in the list (1st 100%, 2nd 85%, 3rd 70%, 4th and 5th 50%, 6th or lower 30%) | 25 |
| Engines that list you | 15 |
| Answers that recommend you and also link to your site | 10 |
Site GEO readiness (40%)
We review the home page and up to 4 key pages, robots.txt, the sitemap and your server’s real response to AI agents.
| Pillar | Points | What it checks |
|---|---|---|
| AI access to your site (crawler access) | 20 | 16 bots in robots.txt and firewall blocks |
| Business details for AI (structured data) | 20 | JSON-LD, business profile, sameAs, FAQPage, reviews, Open Graph |
| Easy-to-quote content | 20 | Figures, questions, lists and tables, authorship, external sources, freshness |
| Text AI can read (without relying on JavaScript) | 15 | Text in the initial HTML, H1, subheadings |
| Trust signals | 15 | About page, phone and address, external profiles, reviews, language |
| Technical basics | 10 | HTTPS, title and description, canonical, sitemap, language, llms.txt |
Where the weights come from
The weights draw on the GEO study (KDD 2024) on figures, quotes and sources; on the Ahrefs study that found brand mentions are the signal most correlated with visibility in AI Overviews; on Ahrefs data about the freshness of cited content; and on the SE Ranking analysis that found no effect from llms.txt, which is why it counts for only 1 point.
Limitations
- AI answers change over time and from person to person. One check is a snapshot; weekly monitoring shows the trend.
- We measure through the official APIs, which can differ from what a signed-in user sees in each engine’s app.
- The technical review covers the home page and up to 4 key pages, not the whole site.
FAQ
Which engines do you query?
ChatGPT, Gemini and Perplexity on the free plan; Claude and Meta AI on the Pro and Agency plans.
Why does my score change between checks?
Because engine answers vary. That is why we use several questions per engine and recommend watching the weekly trend.