Skip to content
Dispensary Rank
Free tools
MSite Check AAI Readability CCompliance Checker VMenu Visibility GGBP Risk Checker LLocation Pages SSchema Generator RReconsideration Builder PSERP Preview KCannibalization Mapper All tools
Free tool, no signup

AI answer readability: what an assistant can read on your page

Ten checks against the HTML with no JavaScript executed, because that is what the models behind AI answers receive. It reports whether there is enough on the page to answer a question with, and enough to attribute it to you.

No signup. The page is fetched once, with no JavaScript executed, which is the whole point of the measurement.

Understanding the check

01What the ten checks look for

Readable text, because a model can only cite what it can read. Address, phone and hours in the text rather than in an image or a widget, because those are the three things somebody asks before visiting. Structured data describing the business, which gives unambiguous facts rather than sentences to interpret. Questions answered on the page, because an answer is the shape an assistant produces. The share of the page that is script. And whether robots.txt allows the crawlers that feed AI answers at all.

02Why this is harder than ranking

A search result is a list. An assistant answer is a paragraph naming two or three businesses. Being fourth in a list still brings some traffic. Being fourth in an answer brings none, because the answer stopped at three.

That raises the cost of a page that is technically fine but thin. Google will rank a thin page somewhere. An assistant will simply use a fuller one instead, and you will never see the impression.

“

Google renders JavaScript late. The models behind AI answers mostly do not render it at all, which makes an embedded menu invisible twice over.

03How to read your result

Read the individual checks rather than the score. The score exists to order work, and it is not a number anyone else recognises. A red mark on the facts check matters more than a red mark on script share, because the first means an assistant cannot state your address and the second only means it has less context.

Run it on a location page and on a menu page. They usually score very differently, and the gap between them is where the work is.

The check that settles it

View source on a menu page, not Inspect Element, and search it for a brand you carry. That is close to what an assistant receives, and it takes ten seconds.

04What to do with what you find

The fixes are the same ones that help ordinary search, which is the useful part: nothing here asks you to build for assistants separately. Put the address, phone and hours in text. Add LocalBusiness markup. Answer real questions on the page rather than describing the business. Get the menu into the HTML.

The one genuinely separate decision is whether to allow the AI crawlers at all. For a publisher whose content is the product, blocking them is defensible. For a dispensary that wants to be named when somebody asks where to buy locally, blocking them removes the only route to being named.

From the accounts

On the accounts I have checked, homepages usually score well and menu pages usually score close to zero. It is the same iframe problem that keeps products out of the index, showing up in a second place, and it tends to be the finding that finally gets the menu fixed after years of being tolerated.

Questions

Questions about AI readability

It is the same measurement taken further. Google renders JavaScript on a second pass, late and unreliably. The models behind AI answers largely do not render it at all, so anything a script draws is simply absent from what they read. A page can pass a normal crawl check and still give an assistant nothing.
Honestly, less than search does today, and nobody has reliable numbers. What is clear is the direction: people increasingly ask an assistant where to buy something locally, and the answer names two or three businesses rather than ten. Being one of the named is worth more than being the fourth result.
That is a business decision rather than an SEO one. Blocking them protects your content from being used without attribution. It also means those assistants cannot cite you when somebody asks where to buy locally. For a publisher the first matters more; for a dispensary the second usually does.
Because the homepage is written into the HTML and the menu arrives through an embed. That gap is the finding rather than a fault in the tool, and it is the same reason the menu does not rank for product searches.
It counts how many of ten checks failed or came back thin. It is a way of ordering work, not a metric anyone else recognises. Read the individual checks rather than the number.
Six an hour from one address. The limit exists so the tool stays available and so it cannot be pointed at somebody else at volume.

What to look at next, and why

Dispensary Menu Visibility Checker why an embedded menu is invisible twice over Dispensary Schema Generator markup that gives an assistant facts rather than sentences Dispensary Site Check the same measurement applied to ordinary search

Menu page scoring near zero?

That is the same fault that keeps your products out of search. Send the domain and you will get a written read on what it is costing and what to fix first.

Get Your Free Audit