I scanned a real clinic, fixed two things, and its AI score jumped 11 points
A dental clinic in Pune asked me to look at why AI answer engines like ChatGPT and Perplexity seemed to skip over its site. I scanned it, found two fixable problems, fixed them, and scanned it again three days later. The score went from 81 to 92, grade B to A.
Score timeline (live scans)
The baseline
I ran the clinic's homepage through the same engine behind the free Report Card on this site. Baseline: 81 out of 100, grade B, from an earlier index scan on 15 July. I re-scanned it live the next day and it sat at 83, still a B. Three things were flagged: a failing meta description, a page title warning, and missing FAQ structured data. The category breakdown made it obvious where the score was leaking: AI Crawler Access was a perfect 15 out of 15 and Technical Signals a perfect 25 out of 25, but Structured Data was 18 out of 25 and Content Clarity was 17 out of 25.
What I actually changed
The page title was a mess: over 90 characters, stuffed with the practice name, a service list, and a leftover line of boilerplate. I cut it to about 52 characters that lead with the practice, the city, and the main service. The meta description was close to nothing, a 4-character placeholder, so I wrote one plain sentence describing the clinic and what it treats.
Neither of these touched a single word of the clinic's actual content. No new page, no new paragraph. Just the two fields an answer engine reads first to decide what a page is and whether to trust it.
The re-scan
Two days later I scanned the same homepage again. 92 out of 100, grade A. Content Clarity moved from 17 out of 25 to a full 25 out of 25. That is the whole story of where the 11 points came from: one category, moved by two edits that took an afternoon, not a content strategy.
The lesson: structure beats volume
The instinct with AI visibility is to write more: more pages, more blog posts, more FAQs. Here, two structural edits (a title and a meta description) did more in three days than months of extra content would have. Structured data is usually the next lever, and it is still open on this clinic's site: their patients ask real questions and the answers exist in the page copy, just not wrapped as FAQPage schema an engine can lift cleanly. If you want to see the kind of questions people are actually asking AI in this category, the questions index is a good place to start.
What is still open
I am not going to pretend this is finished. Structured Data still sits at 18 out of 25 because that FAQ schema has not shipped yet. An A starts at 85, and 92 is comfortably past that, but it is a checkpoint, not a finish line.
Frequently asked questions
How fast can a site's AI score really improve?
This one moved 11 points in three days, but that was because the underlying site was already decent (a clean 25 out of 25 on Technical Signals from day one). Fast fixes work fast when the foundation is already sound; a site with deeper problems takes longer.
Is a title and meta description really enough?
Here, yes, but only because everything else was already in reasonable shape. Those two fields closed the Content Clarity gap because they were the two things actively failing. On a messier site they would help, but they would not be the whole fix.
What is usually the single highest impact fix?
In my experience it is structured data, specifically FAQPage schema for any business that answers common questions in its copy. It turns prose an engine has to guess at into a labeled block it can lift directly, and it is the one item still open on this very clinic's site.
How do I check my own site?
Run it through the free Report Card on this site. It uses the same deterministic engine as this case study, takes under a minute, and tells you exactly what is failing.
The FAQ schema is next; I will scan again once it ships.