When a family asks ChatGPT or Google’s AI “best memory care in our city,” the assistant names a handful of communities. We wanted to know: how often is any given community one of them?
So we measured it. We took 35 mid-market U.S. senior-living operators, ran each one’s flagship community through the question families actually ask — “best [care type] in [city]” — several times across AI engines, and recorded how often the community was named. Not a “rank” (AI answers vary too much for that to be real) — a visibility rate: the share of runs in which the AI recommended them.
The picture is stark.
Finding 1 — Nearly 1 in 4 communities is invisible
8 of the 35 communities (23%) were named 0% of the time. When families ask AI for a recommendation in their market, these communities simply never come up. Their website, their reviews, their reputation — none of it reaches the family, because the family never sees a link. They ask, the AI answers, and a competitor’s name is the answer.
Most operators have no idea this is happening. It doesn’t show up in their analytics, because there was never a click to measure.
Finding 2 — AI visibility is winner-take-most
The communities that do show up tend to show up constantly. Across the 35, the distribution was strikingly bimodal: a large cluster at or near 0%, another large cluster at or near 100%, and relatively little in between. 29% were named in essentially every run.
In other words, AI doesn’t spread recommendations evenly. For a given market it tends to converge on the same few communities and name them again and again. If you’re in that set, you’re compounding; if you’re not, you’re invisible — and there’s not much middle ground.
The average community landed around a 53% visibility rate — named about half the time — but the average hides the real story, which is the gap between the visible and the invisible.
What it means for operators
Three things follow:
- AI is becoming the new front door, and it’s a narrow one. The shift from “ten blue links” to “one synthesized answer” means visibility is more concentrated than it ever was in classic search. Being on page one isn’t the bar anymore; being in the answer is.
- An AI-named family usually comes direct. When the AI recommends you and the family reaches out, that inquiry typically arrives without a referral-aggregator fee attached — the same family that might otherwise cost you several thousand dollars through a third-party referral.
- You can’t fix what you can’t see. The first step isn’t a campaign; it’s a measurement. Each engine pulls from different sources, so a single “AI score” hides where you’re actually losing families. You have to measure per engine, honestly, before you spend a dollar changing anything.
How visible is your community?
This study is a 35-operator snapshot — your own number is the one that matters. Run the free check on this site: enter your community and city, and you’ll see how often each AI engine names you, per engine, with confidence bands. It’s the exact measurement behind this report, and it’s free.
Methodology, stated plainly
- Sample: 35 mid-market U.S. senior-living operators, one flagship community each. A snapshot, not a census of the whole industry.
- Measurement: each community’s market question (“best [care] in [city]”) asked several times across intents; we report the share of runs in which the community was named, per engine, with 95% Wilson confidence bands. Not a rank — AI answers are non-deterministic, so a fixed ranking doesn’t exist.
- Engine coverage: Perplexity — full coverage (all 35), which is the basis for the headline numbers. Google AI Overview — partial, because it doesn’t render for every query and our measurement quota was limited mid-study. We report the engine with complete coverage and flag the gap rather than paper over it.
- Per-operator results are kept private. We share each community’s individual result with that operator directly, not in this public report.
Next edition: a larger sample and full Google AI Overview coverage. Want your community included, or the methodology in detail? Run your free check above — it starts the same conversation.