How we measure SEO, GEO & AI visibility
Primenza scores websites for classic search (SEO), readiness for AI answers (GEO), and AI visibility where your plan allows. This page explains what we look at, how scores are built, and what they do not guarantee — such as a specific Google ranking or an AI citation.
What we measure
We check whether search engines and AI systems can crawl and understand your site: technical and on-page SEO, GEO signals (clear writing, structure, citation-ready content, entities, trust), and optional AI visibility runs when modules and providers are enabled.
SEO scores
SEO audits use crawled HTML and site structure: titles, headings, links, common technical issues, and configured rules. The score summarizes what we found on the pages we measured. It is not a promise of ranking or traffic.
GEO scores
GEO scores show how well content may work in AI answers. We group checks into four areas: readability, citation readiness, structure (entities and hierarchy), and trust. Each page contributes; pillar scores roll up from those measurements.
AI search visibility
AI visibility tracks whether your brand or domain shows up in discovery, mentions, or citations in supported flows — including simulations and, on eligible plans, Market Intelligence runs. Primenza does not control what any AI engine says or cites.
Citations vs mentions
A citation links to your URL or domain in an AI answer. A mention is your brand name without a link. We record these when seen in supported runs; one answer is never treated as the full picture — we store repeats, time, and provider context when available.
Four types of data in the product
- Measured directly
- Taken from a live page fetch, crawl, HTTP response, or a stored AI provider answer with a timestamp.
- Calculated
- Built from documented formulas using measured inputs — for example pillar averages or visibility rates across several runs.
- Simulated
- From a controlled simulation or heuristic when live provider access is off or the product runs a simulation module.
- Inferred
- A careful conclusion from available text — for example mention context parsed from an answer, not a score the provider claims.
What 0–100 scores mean
Scores are on a 0–100 scale so you can compare pages and prioritize fixes inside Primenza. For example, 83/100 means strong readiness on what we measured — not a guarantee of rankings, traffic, or AI citations.
Where data comes from
Primenza crawls, HTTP fetches, HTML analysis, your domain settings, catalog data, and plan-gated provider APIs when configured. Market Intelligence runs also store engine, prompt, time, and algorithm version when applicable.
Good to know
AI answers change often. Results depend on plan limits, crawl depth, sampling, and provider availability. A homepage check is not the same as a full-site audit. We cannot see every query or every index state.
When scores update
Scores change when you run or schedule a new audit, crawl, or visibility run. Past results stay in your workspace with timestamps. Public marketing counters may use display offsets and are separate from your domain scores.
Research & guides
We publish a small set of reference guides and research notes with clear scope and limits — documentation of how we measure, not generic marketing articles.
Frequently asked questions
Does a high GEO score guarantee AI citations?
No. GEO shows readiness on audited pages. AI engines choose sources on their own.
Are AI visibility percentages exact?
No. They are estimates from samples and prompts — use them as trends, not exact market share.
What is measurement_algorithm_version?
In Market Intelligence, the version of our logic used to compute visibility from answer text — separate from provider-reported scores.
Can I verify this in the dashboard?
Yes. Run audits, open per-metric details, and review Market Intelligence history with timestamps where enabled.