A readiness score is a website checklist, not a citation forecast. Here is what goes into it, what a 78 means, and what it will never tell you.
You open a scan report and see 78. Is that good? Does it mean ChatGPT will name you next week? Will a 90 guarantee citations?
No. An AI search readiness score is a website checklist score. It measures whether machines can find and understand your brand on your site. It does not forecast rankings, citation share, or assistant answers.
That distinction matters more than the number itself. Treat the score as a punch list with a grade, not a crystal ball.
What a readiness score actually measures
A readiness score looks at on-site conditions that answer engines and AI crawlers need before they can use you as a source. Think access, identity, and extractable facts — not popularity.
Typical buckets look like this:
| Bucket | What it checks | Why it matters |
|---|---|---|
| Crawler access | robots.txt, bot allow/deny, hard blocks (403/WAF) | If GPTBot or PerplexityBot cannot fetch the page, you are invisible to that pipeline |
| Renderability | Meaningful HTML without waiting on client JS | Many AI fetchers do not run your React app; empty shells read as empty brands |
| Entity clarity | Consistent brand name, Organization schema, About signals | Assistants need one clear "who is this" before they cite you |
| Copy-ready facts | Descriptions, sameAs links, contact/identity fields | Machines prefer structured, stable facts over marketing slogans |
| Technical hygiene | Canonicals, soft 404s, broken schema JSON | Conflicting URLs and invalid markup waste the crawl you did allow |
BrandKnown's readiness score is explicitly website checks, not AI rankings or citation likelihood. A high score means the site is easier for machines to ingest. It does not mean you will appear in answers tomorrow.
What a 78 means (and what it does not)
Say Northstar Analytics scores 78. In plain English that usually means: crawlers can reach the important pages, the brand name is mostly consistent, and schema is present but incomplete or slightly messy. Maybe sameAs is missing Crunchbase. Maybe the homepage still hydrates the company description only in JavaScript. Maybe FAQ schema is broken JSON.
A 78 is useful. It is not a verdict on market position.
What a readiness score never tells you:
- Whether ChatGPT will mention you for "best B2B analytics tools"
- How often Perplexity will cite your blog
- Your share of Google AI Overview citations
- Competitor "authority" in any absolute sense
Pew Research (March 2025 Google browsing study) found AI Overviews on roughly 18% of searches, with traditional-result clicks at 8% when a summary was present versus 15% without. Those are user-behavior numbers in Search — they are not what a site readiness score measures. Do not glue those worlds together in a client deck.
How scores are usually built (without the mystique)
Most useful scores weight pass/fail checks, then roll them into a 0–100. The honest version of the math is boring:
- Run automated checks against the live site (robots, headers, HTML, JSON-LD).
- Mark each check pass, warn, or fail.
- Weight access failures harder than polish issues (a 403 for answer crawlers beats a missing
logofield). - Aggregate into a single number for triage — then show the underlying findings so you can act.
If the product only shows a number and hides the findings, the number is theater. The findings are the product.
How to read your score in practice
Use this decision table instead of chasing vanity points:
| Score band (hypothetical) | Practical read | First move |
|---|---|---|
| Under ~50 | Something fundamental is broken: bots blocked, empty HTML, no entity signal | Fix access and homepage identity before writing more content |
| ~50–75 | Machines can reach you, but identity or structure is muddy | Organization schema, About page, name consistency, sameAs |
| ~75–90 | Solid baseline; remaining issues are specific | Fix named findings; stop re-platforming for a +3 |
| 90+ | Website checks are largely green | Shift effort to off-site evidence and measurement — the score will not move much |
Those bands are a reading aid, not BrandKnown thresholds published as law. Your report's individual findings always beat the band label.
What moves the score (and what does not)
Usually moves it: allowing the right AI user-agents, serving real HTML for brand pages, adding valid Organization (or LocalBusiness) JSON-LD, aligning name across title/H1/schema, fixing invalid JSON-LD, removing soft 404s on key URLs.
Usually does not: publishing ten more blog posts, buying links, tweaking meta keywords (please do not), renaming the color of your CTA, waiting for "the algorithm" to notice your vibe.
SparkToro / Similarweb (Jan–Apr 2026) estimated roughly 68% of Google searches were zero-click. That environment makes on-site clarity more valuable — when fewer people click through, the pages that do get retrieved need to be unambiguous. Still: zero-click rate is industry context, not a term in your readiness formula.
Honest ceilings
You cannot control:
- Model training cutoffs and memory
- Which third-party review site an assistant prefers this month
- Whether Google shows an AI Overview on a given query (Pew's ~18% figure is aggregate, not your keyword list)
- Citation politics between publishers
You can control whether your site is fetchable, named clearly, and structured enough that a crawler does not have to guess.
How to verify improvement
- Note the score and the failed checks before you change anything.
- Ship one cluster of fixes (for example: robots + WAF allowlist for answer crawlers).
- Re-scan the same URLs.
- Confirm the named checks flipped; treat the score delta as a summary, not the proof.
- Separately, if you care about traffic, watch AI referrals in analytics — Similarweb / TechCrunch (June 2025) reported AI platforms sending about 1.13B referrals to the top 1,000 sites (up 357% year over year), with ChatGPT accounting for more than 80% of those AI referrals, while Google Search still sent roughly 191B in the same month. Referral lift and readiness score are related only loosely.
A BrandKnown scan runs in about 60 seconds and is built around these website checks. The free first scan is enough to see whether your "78" is blocked bots, muddy entity markup, or both. Agency plans ($49/mo) exist for teams that re-scan often — the score is still a checklist, just a faster one.
Use the number to prioritize work. Do not use it to promise citations.