(01)What is a GEO score?
A GEO score measures how likely an AI answer engine is to read, understand and cite a page. Where an SEO score predicts a ranking, a GEO score predicts a citation — the sentence ChatGPT, Claude, Perplexity or Google's AI Overviews says back to your customer.
It runs from 0 to 100 across six dimensions. This free scan measures the four it can observe by fetching your page.
| Dimension | Weight | What it measures |
|---|---|---|
| AI Citability | 25% | How quotable your content is when an engine composes an answer |
| Content E-E-A-T | 20% | Experience, expertise, authority, trust |
| Brand Authority | 20% | Third-party mentions on the sources engines trust |
| Technical GEO | 15% | Whether AI crawlers can reach and read the page |
| Schema & Structured Data | 10% | Machine-readable facts about who you are |
| Platform Presence | 10% | Whether the engines cite you today |
(02)How is the score calculated?
We fetch three things — the page, its robots.txt and its llms.txt — then run 21 deterministic checks against them. Each check carries a fixed weight and returns evidence, not an opinion.
No language model scores your page. That is deliberate: a score that drifts between runs cannot be defended to a client, and cannot prove that last month's work moved it. The same URL always produces the same number.
(03)Why does the free scan cover only 70%?
Because two dimensions cannot be observed by fetching your page, and we would rather leave them blank than guess.
- Brand Authority (20%) — depends on how often you are mentioned on Wikipedia, Reddit, YouTube and LinkedIn. Engines lean on those sources to decide you are a real entity worth naming.
- Platform Presence (10%) — requires actually asking ChatGPT, Perplexity, Gemini and Copilot what they say about you today, then reading the citations.
Both are part of the paid audit. Neither can be faked in five seconds.
(04)What is a good GEO score?
Above 75 is good — the page is readable, structured and attributed, and engines have something to quote. Between 60 and 74 is fair. Below 40 usually means one of three things: crawlers are blocked, the content only exists after JavaScript runs, or nothing on the page tells an engine who published it.
Most sites we scan land between 40 and 65. The gap is rarely effort — it is that nobody has told them what the engines are actually reading.
(05)Questions we get
Do you store the URLs people submit?
No. The URL is fetched, scored and returned. Nothing about the scan is written to a database. Results are cached for ten minutes at the edge so repeat scans of the same address come back instantly.
Will this slow down or affect my site?
No. We request one page plus two small text files, identify ourselves as KiwaGEOBot, and give up after eight seconds. It is one visit.
Why did my single-page app score so badly?
Because most AI crawlers do not run JavaScript. If your text only appears after React or Vue mounts, the engines see an empty container. Server-render or pre-render the page and the same content becomes citable.
The score seems harsh. Is it wrong?
Every check quotes its evidence — what it found, and where. If you disagree with a finding, you can check it against your own HTML in a minute. That is the point of scoring deterministically rather than asking a model for a vibe.
A score is a diagnosis. We do the surgery.
Request a full GEO audit