References
The GEO score
Eight weighted criteria, a score out of 10, and a benchmark to read it against.

The short answer
The GEO score rates your site on eight weighted criteria, each marked out of 10, the final score being the weighted average. Three content-access criteria carry half the score on their own. This score is Sekoia’s own indicator: its weights are our internal scale, not official search-engine factors (Google, for one, states that no special markup is required to appear in its AI features). The scale runs from 0 to 10, and a score is always read against your sector's benchmark.
8 criteria
Weighted from 8% to 20%, each marked out of 10, for a total of 100%
50%
Share of the score carried by the three access criteria: structured data, extraction, question-answer
x12
Growth in generative AI referral traffic to retail sites, x17 to travel (Adobe Digital Insights, Adobe Analytics, July 2024 to February 2025, published May 23, 2025)
What the score measures
The GEO score is the deliverable of the second phase of our GEO method, the technical site analysis, which takes 4 to 6 hours. It covers what your brand offers generative engines. It does not yet say whether ChatGPT or Perplexity cite your brand, that is what the visibility tests are for. It is part of every one of our AI search engine optimization engagements.
The access criteria
Three criteria cover how readily an AI can read and extract your content, and together they make up half the score. Structured data carries the most weight, 20%: your JSON-LD schemas, Organization, Article, Service, FAQ, and their consistency from page to page. Extraction ease counts for 15%: clean HTML, a heading hierarchy that holds, readable lists. Voice and question-answer optimization counts for 15%: questions asked the way people ask them, answers given in one or two sentences. If those three are low, we recommend fixing them before the rest.

What builds authority
Three more criteria make up 30% of the score, and they are built over time. Thematic depth counts for 12%: the real coverage of your topics, your pillar pages, your internal linking. Authority signals count for 10%: the data you publish and others pick up, the references you cite. E-E-A-T credibility counts for 8%: named authors, verifiable expertise, a Person schema that actually exists.

What happens off your site
The last two criteria make up 20%, and this is the part most audits leave out. Multi-source presence counts for 10%: Wikipedia, LinkedIn, sector directories, forums, podcasts. A generative engine does not cite you because your site is good, it cites you because you appear in what it has read, and most of that sits outside your domain. Freshness counts for 10%: publishing cadence and updates, because dated content gets picked up less readily.

Our approach in 4 steps
- 01
Assessment
We mark your site on each of the eight criteria, out of 10, with the verified finding behind every mark.
- 02
Weighting
Each mark takes the weight of its criterion, from 20% for structured data to 8% for E-E-A-T credibility. The weights total 100%.
- 03
Score
The final score is the weighted average of the eight marks, out of 10.
- 04
Reading
The score is read against your sector's benchmark, then feeds the gap analysis in our method.
Frequently asked questions
How do you calculate the GEO score?
We mark your site out of 10 on each of the eight criteria, then compute a weighted average. The weights are: structured data 20%, extraction ease 15%, voice and question-answer optimization 15%, thematic depth 12%, authority signals 10%, multi-source presence 10%, freshness 10%, E-E-A-T credibility 8%. They total 100%. Weights are adjusted when a sector calls for it, and the adjustment is then stated in the report.
Which criterion weighs the most?
Structured data, at 20% of the score. But the real lesson lies elsewhere: the three content-access criteria, structured data, extraction ease and question-answer format, carry half the score between them. As long as they are low, we recommend fixing them first.
What counts as a good GEO score?
It depends on your sector, and that is the whole point of the benchmark. The scale runs from 0 to 10: from 0 to 2 the site is invisible to AI, 2 to 4 weak, 4 to 6 average, 6 to 8 good, 8 to 10 excellent. For an agency or a B2B professional services firm, the benchmark we observe sits between 3 and 5 out of 10. A score of 5.1 is not good in absolute terms; it is above that sector's benchmark, which is something you can act on.
Where does my sector's benchmark come from?
From our audits. Every engagement measured with the same scale adds to the benchmark, by sector and size profile. That is why the protocol does not change from one audit to the next: a stable scale is what makes scores comparable to each other and over time.
Does the GEO score say whether ChatGPT talks about us?
No, and that distinction matters. The score rates what your brand offers AI. Whether your brand appears in the answers of ChatGPT, Claude, Perplexity and Gemini is measured separately, through visibility tests, in another phase of the method. One says what you offer, the other what the engines do with it.
How long does it take to get the score?
The technical site analysis that produces the score takes 4 to 6 hours. Diagnosing your GEO score is also the second step of how we start working together, right after the free 30-minute AI visibility audit.
Let's talk about your score
Thirty minutes to find your next step. You describe your situation, we tell you where to start.
- An answer by the next business day
- A person from our team, not a bot
- A first concrete recommendation
Your request has arrived.
A senior team member will get back to you by the next business day. A confirmation email is on its way.