Score Your Cafe's AI Visibility With a Twenty Point Self Test

Thomas Garner

Oct 7

11 min read

You want to know where your cafe stands in AI search without buying a tool or hiring anyone to check for you. This test gives you that in about twenty minutes.

Short answer: twenty yes or no items, five in each of four groups, presence, data, content, and tracking. Each one is observable in under a minute, no tools beyond your phone and a browser. Add up your yes answers for a score out of twenty, read it plainly against the bands below, and pick your three fixes from whichever items failed.

This score is a self assessment, not a national benchmark, and it must never be mixed with Hillcane's formal visibility audit score out of forty or with the Southeast audit's score bands. Those are separate scales built with a separate method for a separate purpose.

Group One: Presence (Items 1 Through 5)

1. Ask an assistant you use, in your own words, something close to is there a good independent coffee shop in my city, and check whether your cafe is named. 2. Ask a follow up, close to what makes {your cafe} different, and check whether it can say anything specific. 3. Search your cafe's name plus your city in a regular search engine and check whether an AI style summary or overview appears above the normal results. 4. If an AI overview or summary does appear for that search, check whether your cafe is named within it. 5. Check whether your cafe's own website shows up anywhere on the first page of that same search.

This group is the hardest to score perfectly on, and that is expected. A newer or smaller cafe may reasonably fail two or three of these five items simply from being less established than a competitor, not from doing anything wrong. Score it honestly anyway; the point of this group is a baseline, not a judgment.

It helps to phrase the questions you ask an assistant the way a real customer actually would, not the way you might phrase a formal survey question. A visitor is more likely to ask something close to where should I get coffee around here than to ask a precisely worded discovery query, and the two phrasings can return different answers. Try a couple of natural variations if the first one comes back with nothing useful, and note which wording you used.

Why These Four Groups and Not Others

Presence, data, content, and tracking were chosen because each one answers a different, plain question an owner can act on: am I showing up at all, is my basic information right, does my site have anything worth pointing to, and am I checking any of this regularly. Other possible groupings, splitting content by page type or splitting data by platform, would add complexity without adding much an owner could act on differently.

Keeping the list at twenty, five per group, was a deliberate choice to keep the whole test runnable in one sitting without it turning into a project. A longer list would likely capture more nuance, but it would also be the kind of checklist that gets started once and never finished, which defeats the purpose of a test meant to be rerun regularly.

Group Two: Data (Items 6 Through 10)

6. Do your Google Business Profile hours match your website's hours exactly, including any holiday hours you list? 7. Is your Google Business Profile verified, shown by a checkmark on the listing? 8. Does your website list a full street address, not just a neighborhood or city name? 9. Is your primary category on your Google profile the accurate, specific one, coffee shop or cafe rather than a broader catch all category? 10. If your site uses structured data or schema markup, is it present at all, even in a basic form?

Google's own guidance on structured data, part of its Search Central documentation accessed September 21, 2026, explains that structured data helps systems understand what a page is about in a consistent, machine readable way. This group checks the raw material an AI system, or a normal search result, would draw from to describe your cafe accurately; if the underlying data is wrong or missing, an accurate summary anywhere downstream becomes less likely.

Across the nationwide census of more than 1,000 independent specialty coffee shops in six regions, snapshot dated September 15, 2026, schema markup status broke down as 368 shops absent, 367 partial and 337 present, with 37 left unclassified, figures rechecked September 21, 2026. That context is here to normalize, not to shame; a large share of cafes nationwide have incomplete or absent markup, so item 10 failing puts you in common company, not an unusual position.

Group Three: Content (Items 11 Through 15)

11. Does your website have at least one blog post or article, however short? 12. Does that content, or any page on your site, answer a real question a customer might ask, rather than only describing your cafe in general terms? 13. Is there a visible date, an author name, or any sign of when a page was last updated, anywhere on your site? 14. Does your about page or homepage include at least one specific, concrete detail that could not apply to any other cafe, a roaster relationship, a neighborhood landmark, a specific house drink? 15. Is there an FAQ section anywhere on your site, even a short one?

In the same nationwide census, 926 of 1,109 shops had zero blog posts, dated September 15, 2026. That single fact explains why item 11 is worth checking honestly rather than assuming it is fine; a large majority of cafes nationwide have not published anything at all, so a single first post moves you out of the largest, most common gap on this whole list.

This group is also where a page becomes something a system trying to extract a clean answer can use. A page with dates, specific details, and a real question answered plainly gives more for a system to work with than a page of pure general description, regardless of what any single assistant ultimately does with that page.

Group Four: Tracking (Items 16 Through 20)

16. Have you checked, at least once in the past month, what an assistant says when asked about your cafe? 17. Do you have any written record, even a simple note, of what was said the last time you checked? 18. Do you know whether your Google profile hours currently match your website, without having to go check right now? 19. Have you looked at your website's search performance data, in whatever form you have access to it, in the past month? 20. Is there a recurring day or reminder on your calendar for checking any of the above again?

This group is different in kind from the first three. It does not check a fact about your cafe's data or content; it checks whether you have a habit of looking at all. A cafe can score well on presence, data, and content and still drift out of date within a few months if nobody is checking. This group exists to catch that drift before it becomes a real gap in the other three.

Scoring and Reading Your Result

Add up all twenty yes answers for a score out of twenty. This is a first draft scoring scheme, built for this self test specifically, not derived from any external study of what a good or bad AI visibility score looks like nationally; we are stating that plainly rather than dressing up a new scale as an established one.

As a plain reading, not a formal band system: a score under ten suggests the basics, particularly in the data and content groups, need attention before anything else. A score in the low teens suggests a reasonable foundation with real gaps still worth closing. A score of sixteen or higher suggests you are mostly maintaining rather than building from scratch, and your remaining work is likely in the tracking group, since that is the one most often neglected once the other three look solid.

Do not compare this score to Hillcane's own formal visibility audit, which uses a different method scored out of forty across eight separate dimensions, and do not compare it to the Southeast audit's score bands, Invisible, Brochure, Trying, Methodical, which come from yet another separate method entirely. Mixing any of these three scales produces a number that means nothing, because none of them were built to be combined.

Choosing Your Three Fixes

Pull your three fixes from whichever items failed, not from a fixed priority order this time, since the four groups here are closer to equally weighted than in some other checklists. That said, a reasonable default when several items fail across groups is to fix data group failures first, since inaccurate hours or a profile that never finished verification can undercut everything else on this list regardless of how good your content is.

If your failures cluster mostly in content, your fastest fix is usually publishing one real post or one real FAQ section, which can move two or three items from no to yes in a single afternoon of writing. If your failures cluster in tracking, your fix is not writing anything at all; it is simply setting a recurring reminder and checking next month.

Whatever three you choose, write them down along with today's date and score. That record is what makes the next run of this test worth something more than a fresh guess.

What This Test Does Not Do

This test does not show you actual AI answers side by side the way a dedicated citation examples post does; it only asks you to check whether your cafe is named, not to catalog full answer text. It also does not track mentions and citations over time the way a dedicated monthly tracking post does; this is a single point in time self test, meant to be rerun, not a running log by itself.

It also, deliberately, makes no promise that a higher score produces more mentions, more citations, or more customers. There is no data available to us that would let us state a reliable link between this specific twenty point score and any outcome like that, and claiming one would be inventing a connection the evidence does not support. Treat a higher score as a sign your basics are in better shape, full stop, not as a predictor of anything downstream.

Making It a Recurring Habit, Not a One Time Project

The real value of this test is not the first score you get. It is the second and third scores, run a few months apart, that let you see whether the fixes you made mattered. Write your score, today's date, and your three chosen fixes somewhere you will find again, a note on your phone, a line in a spreadsheet, anything that survives longer than your memory of today.

When you rerun it, resist the urge to skip straight to the items you already fixed last time to confirm they still pass. Run the whole twenty again in order. Details drift quietly, a corrected hour can slip back out of sync after a schedule change, a published post can be the only one for months while the rest of the content group stays thin, and the only way to catch that drift is checking everything again, not just the parts you remember improving.

Frequently Asked Questions

How do I score my AI visibility?

Run the twenty item self test in this post, five items each across presence, data, content, and tracking groups, all observable in under a minute using only your phone and a browser. Add up your yes answers for a score out of twenty, read it against the plain bands in this post, and choose your three fixes from whichever items failed.

What is a good score out of twenty?

As a plain reading rather than a formal band system: under ten suggests basics need attention first, particularly in data and content. Low teens suggests a reasonable foundation with real remaining gaps. Sixteen or higher suggests mostly maintenance work, often concentrated in the tracking group. This scale is new to this post and not derived from any external study.

Can I compare this score to Hillcane's audit score out of forty?

No. This twenty point self test uses a different, simpler method than Hillcane's formal visibility audit, which is scored out of forty across eight separate dimensions. It should also never be mixed with the Southeast audit's score bands, Invisible, Brochure, Trying and Methodical. All three are different scales built with different methods for different purposes, and combining them produces a number that means nothing.

How many cafe websites have zero blog posts?

In the nationwide census of more than 1,000 independent specialty coffee shops across six regions, snapshot dated September 15, 2026, 926 of 1,109 shops had zero blog posts. That is why item 11 on this checklist, having at least one post, is worth checking honestly. A large majority of cafes currently fail it, so a single first post moves a shop out of the most common gap on the list.

What does the data group check?

It checks whether the basic facts an AI system or a search result would draw from, your hours, address, verification status, primary category, and any structured data, are accurate and present. Google's own structured data documentation, accessed September 21, 2026, explains that this kind of markup helps systems understand a page consistently; if the underlying data is wrong, an accurate summary downstream is less likely.

What should I fix first if I fail several items?

A reasonable default is fixing data group failures first, since inaccurate hours or a Google profile that never finished verification can undercut everything else on the list. If your failures cluster in content instead, a single new post or FAQ section can move several items from no to yes in one afternoon. If failures cluster in tracking, the fix is simply setting a recurring reminder to check again.

Does a higher score mean I will get more customers or more AI mentions?

No, and this post does not claim that. There is no data available that would support a reliable link between this specific score and any downstream outcome like customer visits or citation frequency. A higher score means your basics are in better shape, which is worth having on its own, not a predictor of anything beyond that.

How is this different from the twenty prompts post?

The twenty prompts post is a repeatable set of test questions to run on ChatGPT and Gemini each month, focused on what specific answers come back. This self test is a broader, one time check across presence, data, content, and tracking, meant to give you a general standing score rather than a detailed log of individual answers.

How often should I run this self test?

Rerun it on a fixed interval you choose, monthly or quarterly, and put that interval on your calendar, which is item 20 on the list. There is no external schedule required. The value comes from comparing your own score over time, so write down each run's date, score and three chosen fixes, then run all twenty items again in order rather than only the ones you fixed last time.

Do I need any paid tools to run this test?

No. Every item is designed to be checked with a phone, a browser, your own Google Business Profile and your own website, all free to access, and each takes under a minute. The whole test runs in about twenty minutes. It intentionally avoids any paid AI visibility tool, since the goal is a baseline any cafe owner can run today and rerun next quarter without buying anything.

Work with Hillcane

Twenty yes or no checks, four groups, no tools required, and a score you can rerun next quarter to see if anything actually moved.

If your score points at gaps you do not have time to close yourself, reach us at hillcane.co/contact or (256) 384-2449.