Which SEO and GEO numbers are worth citing?

This page holds two kinds of numbers, deliberately kept apart. The first are thresholds from primary sources: Core Web Vitals limits, title and description lengths, our score scale. The second are our assessments from working on websites, labelled as assessments rather than as the findings of a study.

Published 2026-09-25 SeoScore

This page exists so a number can be taken and cited without anyone wondering where it came from. Hence two separate lists, never mixed.

A. Thresholds from primary sources

These are not our numbers. Each carries a link to the document that sets it.

Core Web Vitals thresholds (web.dev, Google):

LCP "good" — up to 2.5 s;
INP "good" — up to 200 ms;
CLS "good" — up to 0.1;
the value assessed is the 75th percentile, not the average.

Text limits in a search result:

<title> practically fits about 50–60 characters;
description — about 120–160 characters;
the real limit is pixels, not characters, so the numbers are a target.

Structured data:

the recommended format is JSON-LD;
the data must describe what is visible to a person on the page.

AI bot names, as published by the vendors themselves:

OpenAI — GPTBot, OAI-SearchBot, ChatGPT-User;
Perplexity — PerplexityBot;
Anthropic — ClaudeBot;
Google, for generative products — Google-Extended.

B. Our assessments from ~500 audits

These numbers are not precise statistics. They come from roughly 500 audits run on our earlier platform and from what we see today. The exports from that work are gone, so they are given as ranges.

Cite them exactly that way: SeoScore assessment from ~500 audits, ranges, September 2026.

the usual technical SEO score — 30–60 out of 100;
a freshly launched, tidy site holds about 50 from its template alone;
above 80 is reached by a small minority;
on the AI visibility side the usual score is lower than SEO — typically 30–50;
the most common structured-data state — one Organization node from the template and nothing else;
the most common Article mistake — no author and no dates;
the most common robots.txt state regarding AI bots — not named at all.

How to cite this

If you take a number from part A — cite the primary source, not us. It is right there.

If you take one from part B — label it as an assessment. We care more that a number is correctly marked than that it is attributed to us.

Why so strict: a wrong number carrying our name costs more than an uncited one. We apply the same rule to findings in reports — better "cannot verify" than "all good".

Where these numbers are used

Each topic in full: a typical score, the state of AI search and how the score is calculated.

Your own number comes from a free scan — always more accurate than any average.

What gets checked is listed under what we check.

Sources

Frequently asked

May I use these numbers in my own article?

Yes. Cite part A together with its primary source, and part B with the label "SeoScore assessment" and the date. Because the exports from those audits are gone. A percentage needs a sample you can show; one reconstructed from memory would sound more precise than it is. A range says the same thing without lying about precision.

How often are these numbers updated?

Part B is revisited when what we see in practice changes. Part A when the source itself changes; each link is verified before publishing.

Can the score scale change?

It can, and then its version is raised while old scans keep their own. Otherwise yesterday's number would change for no reason on the site.

Benchmarks →

Check yours

A free scan shows which of the things described here your site already does, and which it does not.

Scan a website →