Direct answer: This is a 20-check, 5-group, 100-point self-audit for corporate website AI readability, covering crawlability, static extractability, structured data, language declarations and inquiry paths. Tick what you can verify in static HTML; the page calculates the total, group scores and prioritized fixes. It sends no data and does not predict AI citations.
The checklist comes from our pilot of 21 Changzhou AI-device companies: among 7 analyzable homepages, none combined a unique H1, structured data and hreflang. Turning that observation into tick boxes lets teams recheck their own site instead of reading another trends piece.
Tick only what you can confirm in static HTML; if unsure, leave it blank.
The score is for internal ordering, not external ranking or vendor scoring.
The audit is not an AI citation, and it does not replace performance, content or index measurement.
Start the audit (local, nothing uploaded)
Priority fixes (unchecked, by priority)
- Tick items to see the remaining work.
Checks and scoring stay in your browser: no cookies, no localStorage, no network requests; reload clears them. Without JavaScript, download the CSV checklist and review manually.
Score formula and grade bands
Total = checked ÷ 20 × 100, rounded to an integer; group score = checked ÷ group total × 100. All 20 checks weigh one point. Equal weights are not a scientific model; they avoid dressing subjective weights up as precision.
| Score | Band | Meaning | Do first |
|---|---|---|---|
| 0–39 | Basics missing | Crawlability or static content has clear gaps | Check robots, status, JS dependence and H1 |
| 40–59 | Needs work | Readable, but structure and inquiry path are incomplete | Add H1, JSON-LD, contact entry |
| 60–79 | Usable | Most basics present, declarations still missing | Add hreflang, privacy, schema consistency |
| 80–100 | Basics complete | Most of the 20 items are verifiable in static HTML | Move to performance, content and index measurement |
The band describes these 20 checks only, not overall website quality, rankings, AI mentions or inquiries.
Why these 20 checks
They come from three places: the gaps we actually observed in the 21-company pilot on 31 October 2026; public rules for robots and structured data; and the items most often missed in website acceptance that can still be verified statically. Image compression, Core Web Vitals, backlinks, index coverage, content quality and AI citations are excluded because they need separate tools and data.
Three choices need explaining: copyright years are not used, because a footer year does not prove an update; an English page is not treated as "good multilingual", only as evidence of a real alternate plus declarations; keyword density is not checked, because no reliable threshold exists and it invites stuffing.
Who it is for, and who it is not for
For: marketing/IT owners at manufacturers, vendors doing a pre-delivery check, buyers screening supplier websites, content teams taking inventory before a redesign. Not for: third-party audit reports, ranking or competitor benchmarking, or measuring AI citation outcomes; those need independent measurement.
How to use it (10 minutes)
- Open the homepage,
/robots.txt,/sitemap.xmland one product page. - View once in incognito and once with JavaScript disabled; tick only what static HTML confirms.
- Tick the 20 items, record total, group scores and pending items; print or screenshot the result.
- Fix priority 1 first, then recheck under the same URL, date and browser conditions.
Limits: what it does not replace
- Not a ranking: the score is valid for this checklist version only; do not compare companies with it externally.
- Not an AI citation predictor: allowed robots and complete schema do not guarantee crawling, indexing or citation.
- Not a performance test: TTFB, Core Web Vitals, image weight and field data are not measured.
- Not a content-quality review: accuracy, originality, readability and conversion copy are out of scope.
- Self-reported: the user ticks the boxes and may over- or under-count; third-party review needs sampling.
Sources and definitions
- RFC 9309, Robots Exclusion Protocol: rfc-editor.org.
- schema.org Organization: schema.org.
- Google Search Central, structured data general guidelines: developers.google.com.
- Pilot behind the checks: AI readability of 21 Changzhou AI-device websites.