What does the free check actually measure?

How often an AI assistant names your community when asked the questions families ask — “best assisted living near [city]”, “top-rated memory care in [city]”, and a conversational version. It is a rate: appearances divided by runs, per engine, with a 95% Wilson confidence band. It is not a position in a list.

Why don't you show an “AI rank”?

Because none exists. The same question returns a different set and order of communities nearly every time it is asked, so any single answer, and any “rank” read off it, is noise. What is stable is how often you appear across many runs. That is what we report, and we show the confidence band so a small sample is visibly a small sample.

Which AI engines are checked?

ChatGPT, Google AI Overviews, and Perplexity, each measured separately, with Gemini following. If an engine returns no usable answers during your scan it is shown as unavailable rather than as a fake number. The paid product adds daily tracking and share of citation versus your competitors on all engines.

How long does it take, and is it really free?

About a minute for a real scan; the page polls automatically. It is free, there is no sign-up, and the result link can be shared with a colleague. The email unlock is optional and only reveals the extra detail.

My community scored 0%. Does that mean AI never recommends us?

It means that across the runs in this scan, on these questions, no engine named you — and the confidence band tells you how sure that is. A 0% with a wide band on a handful of runs is a strong hint, not a verdict. It is also the normal starting point: in our study of 35 operators, nearly one in four were never named. The fix is concrete: give each engine a page it can quote.

We scored 90%. Why do we still need anything?

Two reasons. Visibility moves — engines change what they cite monthly, and a competitor can overtake you without anyone noticing. And a high score only pays off if the inquiries it produces are captured and joined to move-ins in your CRM, so the channel gets credit and budget. That is the ongoing product.

Why can I be named on Perplexity but missing from Google's AI Overview?

Each engine trusts a different set of sources. Google AI Overviews leans on Google Business Profile data and review sites; Perplexity and ChatGPT cite pages they can crawl and quote. Being well described in one set of sources and absent from another is common, which is why every engine is measured on its own.

What happens with the email I enter to unlock the detail?

It is stored with that scan so we can send you the unlocked report and, if you opt in, a re-check later. It is never sold or shared. The full policy is on the privacy page.

Can I run the check for a competitor or a whole portfolio?

Yes, one community at a time, within the fair-use limit of a few scans per minute. Identical scans are served from a 24-hour cache. For a portfolio, the dashboard tracks every community on a schedule and ranks the weakest first.

AI visibility for senior living

Is your community in ChatGPT?

When families ask AI for senior living recommendations, does your community get named? Check across ChatGPT, Google AI Overviews & Perplexity — free.

Community name — use your full name (e.g. "Brookdale Westlake Hills")

City

State opt.

Care type — pick up to 3

Assisted living Memory care Independent living Skilled nursing

Local competitors — optional, comma-separated; lets us show who AI names instead

How the free check works

1

Enter the community, its city, and the care types

Use the community's full name as families would type it (for example “Brookdale Westlake Hills”), the city, and up to three care types. Optionally list the competitors you most want to compare against; only brands you name are matched, never guessed.
2

We ask each AI engine the questions families ask — several times

For every care type the checker asks three family-style questions (“best memory care near Austin”, a conversational version, and “top-rated”), on each engine, two times each. The answers and the pages they cite are parsed for the community's name. Real engines take about a minute; the page polls until the scan is complete.
3

Read the per-engine visibility rate, share the link, unlock the detail

You get the share of answers in which your community was named, per engine, with a 95% confidence band. The result has a shareable link that needs no account. Entering an email unlocks which competitors or aggregators the engines named instead and which third-party sources they cited.

Checks by care type

Families ask by care type, and engines answer by care type, so a community can be visible for assisted living and invisible for its memory-care wing. Each page below shows the questions families ask for that care type, what the engines cite, and the usual gaps, with the check pre-filled.

What the result shows

An overall visibility rate — the share of all AI answers in the scan that named your community, with how many answers that was out of.
A rate per engine, measured separately, each with a 95% confidence band so you can see how much of the number is signal and how much is sample size.
Who the engines named instead (unlocked) — the competitors you listed and the aggregators, such as A Place for Mom or Caring.com, that took the recommendation.
Which sources each engine cited (unlocked) — the third-party pages the answers were built from. This is the fix list in raw form: if you are not on those pages, you are not in the answer.
A shareable link that opens the same result without an account, so the number can go to a colleague, an agency, or a board deck as is.

How we measure, and why there is no “AI rank”

The checker runs the same measurement core as the paid product, on an untracked brand. These are the rules it follows, and they are the same rules the dashboard is held to:

We ask each engine the questions families actually ask — for example “best memory care near Austin” — several times per day, because AI answers change run to run.
A community's visibility rate is the share of those runs in which it was named. It is reported per engine with a 95% Wilson confidence interval, so a small sample shows as a wide band rather than a false-precise number.
We never report an “AI rank.” The same question returns a different ordering nearly every time it is asked, so a rank is noise; a rate over many runs is stable.
Only known brands are matched (the community and the competitors it tracks); we never invent a competitor from free text.
An engine error is excluded from the sample, not counted as a zero. A Google results page without an AI Overview is counted, because the community genuinely could not be named there.
Attribution is directional: every inquiry-to-move-in link carries a match method and a confidence score, and the funnel is labeled accordingly.

The practical consequence: a scan that asks each engine six times per care type is a small sample, and the band will be wide. The number is honest about that. Ongoing tracking in the dashboard narrows the band by asking every day.

Which engines are checked, and how

ChatGPT: Asked through OpenAI's API with web search on; the answer text and the pages it cites are parsed for named communities.
Google AI Overviews: The Google results page is fetched for the same question; the AI Overview block, when Google shows one, is parsed. A results page with no AI Overview counts as a run where you could not be named — that is real signal, not an error.
Perplexity: Asked through Perplexity's Sonar API; the answer and its citations are parsed.
Gemini: Asked through Google's Gemini API with grounding on; the grounded answer and its sources are parsed.

Only engines that returned real answers during your scan are shown. An engine that could not be reached is marked unavailable rather than filled with a made-up figure, and engines are added one at a time as they go live, so “checked on every engine” is never claimed before it is true.

How to read your score

Rate	What it usually means	First move
0–20%	The engines have nothing quotable about you. Aggregators are taking the recommendation — and the referral fee.	Publish a citable community page and fix the listings the engines read. Start with why you're missing from AI Overviews.
20–50%	Named sometimes, usually on one engine. Check the unlocked detail for which engine and which sources.	Work the weakest engine's sources. See GEO for senior living.
50–80%	You are a regular answer. The question becomes how many of those families you capture and convert.	Tag AI-sourced inquiries and join them to the CRM. See track move-ins by source.
80%+	Winner-take-most territory. The risk is losing it quietly when an engine changes what it cites.	Track daily and alert on drops and competitor overtakes; prove the channel's move-ins to protect its budget.

Read the band before the number. A 40% with a band of 15–70% on six runs and a 40% with a band of 35–45% on two hundred runs are different facts; the first says “probably in the middle”, the second says “in the middle”.

What a low score costs

When an AI assistant hands a family to a referral aggregator instead of naming your community, that family usually arrives through the aggregator, and the move-in carries a fee of 50–120% of the first month's rent, commonly $3,500–$12,000. A family that came direct because the AI named you costs nothing. One avoided fee pays for a year of tracking many times over, which is the whole economic case; the detail is in what referral fees actually cost operators. The pattern across operators, including how many are invisible today, is in the State of AI Visibility in Senior Living.

Frequently asked questions

What does the free check actually measure?: How often an AI assistant names your community when asked the questions families ask — “best assisted living near [city]”, “top-rated memory care in [city]”, and a conversational version. It is a rate: appearances divided by runs, per engine, with a 95% Wilson confidence band. It is not a position in a list.
Why don't you show an “AI rank”?: Because none exists. The same question returns a different set and order of communities nearly every time it is asked, so any single answer, and any “rank” read off it, is noise. What is stable is how often you appear across many runs. That is what we report, and we show the confidence band so a small sample is visibly a small sample.
Which AI engines are checked?: ChatGPT, Google AI Overviews, and Perplexity, each measured separately, with Gemini following. If an engine returns no usable answers during your scan it is shown as unavailable rather than as a fake number. The paid product adds daily tracking and share of citation versus your competitors on all engines.
How long does it take, and is it really free?: About a minute for a real scan; the page polls automatically. It is free, there is no sign-up, and the result link can be shared with a colleague. The email unlock is optional and only reveals the extra detail.
My community scored 0%. Does that mean AI never recommends us?: It means that across the runs in this scan, on these questions, no engine named you — and the confidence band tells you how sure that is. A 0% with a wide band on a handful of runs is a strong hint, not a verdict. It is also the normal starting point: in our study of 35 operators, nearly one in four were never named. The fix is concrete: give each engine a page it can quote.
We scored 90%. Why do we still need anything?: Two reasons. Visibility moves — engines change what they cite monthly, and a competitor can overtake you without anyone noticing. And a high score only pays off if the inquiries it produces are captured and joined to move-ins in your CRM, so the channel gets credit and budget. That is the ongoing product.
Why can I be named on Perplexity but missing from Google's AI Overview?: Each engine trusts a different set of sources. Google AI Overviews leans on Google Business Profile data and review sites; Perplexity and ChatGPT cite pages they can crawl and quote. Being well described in one set of sources and absent from another is common, which is why every engine is measured on its own.
What happens with the email I enter to unlock the detail?: It is stored with that scan so we can send you the unlocked report and, if you opt in, a re-check later. It is never sold or shared. The full policy is on the privacy page.
Can I run the check for a competitor or a whole portfolio?: Yes, one community at a time, within the fair-use limit of a few scans per minute. Identical scans are served from a 24-hour cache. For a portfolio, the dashboard tracks every community on a schedule and ranks the weakest first.

Fair use: a few scans per minute per visitor; identical scans are served from a 24-hour cache; the free tool has a daily engine budget and may ask you to leave an email when it is reached. Updated 2026-10-16. Questions about the method? Read how we measure.