What does my AI visibility score mean, and what do I fix first?

Your AI Visibility Score, from 0 to 100, measures how much of your website an AI crawler could read and use on the day of the scan. It covers 5 layers: Front Door, Entity Integrity, AI Readability, Mesh Integrity and Authority Signal. It is not a ranking, and it does not count ChatGPT mentions. If your score is low, fix the Front Door first, since a page the crawler cannot fetch gives the other 4 layers nothing to read, and fix Entity Integrity next.

By Margareta Petrovic, CISA, CISSP, founder of Visibility Mesh. At Visibility Mesh we measure how readable websites are to AI assistants, and we publish what we find. We published this article on October 5, 2026.

You ran the free scan, and an email arrived with a number, a band and 5 category scores. Below we explain each part in plain words and say what to do about it. The later sections tell you which fix to make first.

What the number measures

Our scan loads your key pages the way an AI crawler receives them. It reads the structured data and the visible text. It also checks your robots.txt rules for 8 AI crawlers, one at a time.

Each page earns points across the 5 layers. The total is the share of available points, from 0 to 100.

We leave traffic, rankings and backlinks out on purpose. That is why some famous brands score lower than small shops. A separate AI mentions check records how often assistants name a business, and we never fold it into the score.

Two scans of an unchanged site are rarely identical, and they often land within about 2 points of each other. Our reports treat a smaller move as no measurable change.

What the 4 bands mean

The AI Visibility Maturity Model puts every score in one of 4 bands. We round the score to a whole number first, so the band matches the number on your scorecard.

A low score is the common result

Visibility Mesh has scored 957 websites to date. Of those, 759 form the live benchmark on our current method. Their median score is 43.8 out of 100, and 67% score below 50. No site in the benchmark has reached the Leading band. All of these figures were measured on October 4, 2026.

The 433 online stores in that benchmark score lower still. Their median is 40.6, and 76% of them sit below 50.

So a score in the 40s puts you in the middle of the field. The top 10% of sites in the benchmark score 58.3 or more, and the highest score is 71.5.

The 5 layers in plain words

Your scorecard uses category names. Our published framework uses shorter names for the same 5 layers, and we give both so you can match them.

Front Door, called Technical Foundations on your scorecard

The Front Door asks whether a machine can read the page at all. A bot wall can stop the crawler. So can an error page that reports success, or content that appears only after JavaScript runs. In each case, the page can look fine in your own browser.

Entity Integrity, called Schema and Structured Data

Entity Integrity asks whether your brand and each product are clean, machine readable facts. It checks the structured data for price, availability and identifiers. It also checks that the markup agrees with the page. If the markup lists one price and the page shows another, Entity Integrity flags the conflict.

AI Readability

AI Readability asks whether the answer is stated or buried. A good product page says plainly what the item is, what it is made of and who it suits. An assistant can quote a sentence like that. When the same facts are buried in marketing copy, it usually passes over them.

Mesh Integrity, called Semantic Architecture

Mesh Integrity asks whether your pages form one map. Collections, products and key pages should link to each other, so a machine can see what the brand sells and what each product is for. If nothing links to a page, a machine has trouble finding it or knowing where it belongs.

Authority Signal, called Authority and Operational Discipline

Authority Signal asks whether your identity is consistent, current and backed by evidence. It checks dates and links to your real profiles. It looks for your About and contact pages. It also checks that ratings in the markup match the page.

Which fix to make first

Check the Front Door before anything else. If the crawler cannot fetch a page, the scan has nothing on it to grade in the other 4 layers. For a Starting Point site, no other change gains as much. Polishing product copy on a page the crawler cannot open is careful work for an audience of nobody.

A Developing site should fix Entity Integrity next. In our June 2026 study of 216 Shopify storefronts, Entity Integrity was the weakest layer. Of those stores, 72% failed it.

After that, the work becomes routine. Put AI Readability, Mesh Integrity and Authority Signal on a monthly check, the way an IT team handles patching.

What a low score costs you

An assistant answers from what it can read. Some product pages reach the crawler nearly empty. In that case the assistant has no facts about your product, so it recommends the stores it could read instead.

The score shows whether an assistant can read enough of your site to name you at all. It does not estimate lost sales.

Being readable is one condition among several. Authority and freshness also count, and so does what other sources say about you. How closely a page matches the question matters as well.

What to do this week

When it makes sense to hire help

Steps 2 to 5 of the checklist find problems. Fixing them usually means editing theme files, a firewall rule or the structured data your apps print. Most owners hand that part to a developer.

If your lowest layer is the Front Door, Foundation is the fix we built for it. It covers Shopify and WordPress. If Entity Integrity is low across your catalog, look at Schema Mesh. It includes everything in Foundation and builds structured data from your own products, services and pages.

On Webflow, Wix, Squarespace or a custom build, choose the Full Assessment and Roadmap. It costs $1,490, paid once, and gives your developer every fix in the order we would make it.