NL · FR · EN ceeme · guide

How ceeme scores your AI visibility (and why stricter = more honest)

Methodology · transparent · 2026

ceeme calculates your AI visibility by having an AI answer 5 realistic customer questions live with real web search, measuring for each question whether you are mentioned, in which position, and whether your website is cited as a source. A score is only worth as much as the method behind it — so we explain it exactly, and why our score is deliberately stricter than the headline number of many other tools.

What we measure

ceeme comes up with 5 realistic questions your customers would ask ChatGPT, Gemini or Google AI — without naming your business — and has an AI answer them live with real web search. Per question we look at three things: are you mentioned, in which position, and are you cited as a source with your own website.

The scoring

Each question is worth up to 20 points. Your position in the answer sets the base; citation — the AI uses your site as a source — is the bonus that makes the difference:

Position in the AI answer	Without citation	With citation
1st mention	15	20
2nd mention	13	18
3rd mention	11	16
4th–5th mention	9–7	14–12
Not mentioned	0	—

Your score is the percentage of attainable points, counted across the questions that could actually be checked live. First everywhere but never cited? Then your maximum is 75/100. Only those who are mentioned AND count as a source can reach 100.

Why citation weighs so heavily

Citation is the core of GEO. If an AI names you but never consults your website, your visibility comes from other people's content — reviews, directories, third-party articles. That can shift tomorrow without you being able to do anything about it. If you're cited, you are the source yourself and you're structurally in the answer.

The verdict is stricter than the number alone

Alongside the score, two extra rules guard the honesty of the final verdict:

• Share of voice: if your competitors together get mentioned far more often than you, you never get "dominant" as a verdict — however high your score. Your share of the answers is always shown.
• Full measurement or no measurement: if a live question fails, we retry and insert a replacement question if needed. If the measurement stays incomplete, we show that explicitly (~score) and the scan doesn't count as your free scan.

Why your ceeme score can be lower than elsewhere

Many AI visibility tools show your presence as the headline number: in how many prompts do you appear? Citations sit next to it as a separate figure. The same business can therefore see "100%" there and 75 at ceeme — that's not a worse measurement, but a stricter one: at ceeme, 100 means you truly win, credentials included. We think that's more honest than a high number with a footnote.

And the other scores on your report?

• SEO basics: hard-checked on-page essentials (title, meta description, schema, H1, sitemap, freshness) — no assumptions.
• Technical SEO: speed and code, measured via PageSpeed.
• Digital polish: measurable design signals (colour consistency, typefaces, alt texts, mobile, clutter) from your homepage and stylesheets.

Every report additionally labels each recommendation: ✓ verified in this scan (hard data) or to confirm with a deep scan (assumption). So you always know what's proven and what isn't.