SAVI Report · UK Care Homes · 2026
When a family asks ChatGPT, Gemini or Google's AI to recommend a care home near them, it reads websites and puts a few names forward. We checked 2,307 UK care home websites, including the corporate groups, to see how ready each one is to be picked. Not one is AI-Ready, and the homes the regulator rates Outstanding are no easier to find than the rest.
The plumbing an AI needs to reach a home is fine, and nearly every home has it. Giving the AI clear, well-labelled information, who runs the home, what the services are, what families say, is the part that wins the recommendation, and almost no one does it. The door is open; the shelves are bare.
Every home gets a score from 0 to 100 for how ready it is to be found and recommended by AI. To reach the top "AI-Ready" band you need 80, and not one home gets there. This is the lowest-scoring cohort we have published.
One home in five is Strong and more than half sit in the middle, while 583 homes, one in four, fall below 40 and are effectively invisible to AI. When a family asks an AI to recommend a care home, it names two or three, not ten. When nearly every home looks the same, the few that stand out win far more than their share.
The score is built from eight areas. Homes score well on the plumbing an AI needs to reach them, the technical basics and the crawlers being let in, but fail on everything an AI needs to trust and cite them: structured data, brand authority and the evidence of expertise. The door is open; there is nothing labelled for a machine to lift and repeat.
Of the 2,307 homes, 2,196 carry a current CQC rating. If the regulator's verdict shaped what AI recommends, the Outstanding homes would stand clear of the rest. They do not. The regulator's top badge buys just 3.9 points of AI visibility, Outstanding 50.0 against Requires-improvement 46.1. CQC rating and AI visibility are essentially decoupled.
Not one of the 125 Outstanding-rated homes is AI-Ready. One in five (24 of 125, 19%) is effectively invisible to AI, and 41% of Outstanding homes score below the cohort average. The single most AI-visible care home in the country, Watermoor House (77.8), is rated only "Requires improvement", and it out-scores every Outstanding home. The best-scoring Outstanding home, Loveday & Co, reaches just 75.1, still short of the AI-Ready line.
A family asking ChatGPT or Google's AI to recommend a care home is not shown the best-rated homes. The rating a family would trust is invisible to the tool they now ask.
On our measure, UK care homes sit at the bottom of every sector we have benchmarked. They average about 48 for AI-search readiness, roughly level with law firms (around 47) and behind dentists and accountants (around 52) and aesthetic clinics (around 56). Care homes are the least AI-visible sector we have measured, and not one of the 2,307 clears the AI-Ready bar. The homes that come closest are not doing anything exotic. They do a few specific, cheap things the rest have skipped.
Homes build technically reachable sites: on the technical basics they average around 72 and on AI citability around 71, and the crawlers are let in. But they give AI almost nothing to trust or lift: structured data averages 27.4 and brand authority 33.3, the two weakest areas. In a decision families make on trust, that gap costs them the recommendation.
The homes near the top do four cheap things almost everyone else skips: they name the registered manager in machine-readable form, mark the home up as a LocalBusiness or Organisation with its address and services, publish the real questions families ask as FAQ content, and show fees and CQC information as structured data.
The highest-scoring independent homes in the country, all notified before publication, are Watermoor House (77.8), Valerie Manor Care Home (76.4), Oakland Care (75.4), Loveday & Co (75.1, the only Outstanding-rated home in the group), Abbey Total Care Group (74.2), Trust Care (73.8), Country Lodge Nursing Home (72.2), Minton Care Hotels (71.5) and Nutley Hall (71.2). Every one of them is short of the AI-Ready line, and the gap between them and the rest is a few labels, not the quality of the care.
The same thing keeps happening. Homes are reachable, the crawlers are let in, and the site works, then no one adds the small labels that let AI read and trust who runs the home, what it offers and what families say.
Structured data averages just 27.4 out of 100. Almost no home publishes the machine-readable facts, the address, the services, the fees, that an AI needs to lift and repeat.
Brand authority averages 33.3. Few homes show the credentials and trust signals an AI weighs before it will put a name forward.
Not one home clears the AI-Ready mark. The leader, Watermoor House, reaches 77.8. The recipe is known, and no one yet follows it fully.
583 homes, one in four, score below 40, the point at which an AI can barely use the site at all. The regulator may rate them well; the machine cannot see them.
Choosing who to trust with a relative's care is won on trust and credentials: a named registered manager, a current CQC rating, genuine reviews, real staff. Yet brand authority averages just 33.3 and structured data 27.4 across the index, so the very things a family (and an AI) look for before enquiring are rarely labelled in a way AI can read. This is the open goal.
Check your inbox in the next few minutes for the full report. If it is not there, take a look in your spam folder.
How we did this. We checked each home's live website the same way an AI reader would, looking at more than 250 things that affect whether AI can find, understand and recommend it, and turned that into a score out of 100. The benchmark covers 2,307 UK care homes identified from the CQC register and public listings, each audited in 2026 for AI-search (GEO) visibility. Ratings shown are the home's current CQC rating at the time of audit. We name only the ten highest-scoring homes, all notified before publication; the rest stay anonymous.