AI visibility for vets: named for which animal, which service, and when

Named, recommended or first, for one animal and one service

Three verdicts can be read from every answer, here for a practice; our methodology records two of them, named and first. Named: the practice appears anywhere in the answer, under any form of its name, including an aside such as "another option is [Practice name]". Recommended: the answer puts it forward as a fit for this owner's question, as a practice that sees rabbits, say, or one that is open on Saturdays. First: it is the first practice the owner reads.

Each verdict belongs to one line of the panel, not to the practice. A practice can be first for "cat dental cleaning in [town]" and absent for "vet that sees bearded dragons in [town]" in the same month, and both readings are true. Folded into one number too early, they hide the line that most needs work, so keep the verdicts line by line and read the total afterward.

The total is the visibility score: the share of the twelve lines in which the practice was named, from 0 to 100, with first place reported beside it and no weighting. The method asks each line once per reading, of one assistant, Claude, and the panel file with its base record is in setting up prompt monitoring. What follows is what a veterinary practice adds on top: the brackets, and a few fields in the record.

The wrong answer that costs an owner the most

Wrong answers do not cost the same. An answer that says you see reptiles when you do not costs an owner a phone call. One that repeats last year's price for a spay costs a surprise at the front desk. One that says you are open at 11 p.m. on a Tuesday, when you close at 7, sends an owner with a sick animal to a locked door, and costs them the time it takes to find somewhere open.

That order sets the order of the checks. The hours are read before the verdict, because an answer that names you with the wrong hours leaves the owner worse off than one that leaves you out. It sets the panel too: each animal you want asked about gets a line for the hours where a wrong answer costs most, which are the evening, the weekend and the public holiday.

The page an owner should land on after close is a separate job, set out in the after-hours page for a practice. The question here is narrower: what the answers say about your hours, and whether it is what your own pages say.

A panel bracketed by species, service and hour

A practice's question panel runs to twelve lines, the size we use for paying clients. Build it as a grid: one row for each animal you want to be asked about, a daytime line that names a service, and a line for the same animal outside your hours. Everything in brackets is yours to replace. The lines are templates, not questions we recorded. The first full run is your baseline, and every later month is read against it.

Two lines sit outside the grid. One asks for "vets in [town] accepting new patients for a [puppy]". The other names you on purpose, "[Practice name] hours on [the next public holiday]", and tests what an answer says about your hours to an owner who already has your name.

Swap the rows for the animals you see and want more of. A practice that sees only dogs and cats can drop the exotics rows, or keep one, for the reason in the questions at the end. Keep every line to finding and choosing a practice. A question about what is wrong with an animal is for a veterinarian, and the panel does not measure it.

Why an after-hours line names the hour instead of saying now

An owner at night might type "open now". A panel line run in the afternoon cannot, because "now" asks about the moment the line runs: "emergency vet open now in [town]" run at 2 p.m. tests nothing about 11 p.m. There are two ways around it. Write the hour into the line, as in "open at 11 pm on a tuesday", or run the line at the hour it is about and write down when it ran.

Writing the hour in is the better default, for two reasons. The panel stays fixed, so the line run in March compares with the same line in April. And the hour in the line is an hour your pages state, so what the answer says can be checked against them. If you also want to see what an answer says at the hour itself, add a separate line that says "now", run it at that hour, and log it apart from the rest.

Public holidays get their own treatment. A holiday line is worth running in the weeks before the holiday, once your pages carry the holiday hours; run the rest of the year, it tests a page with nothing to say yet. Date it, run it ahead of each holiday, and compare one holiday with the last.

The fields a practice adds to each answer's record

The base record holds the date, the assistant, the run, the answer pasted in full, and the verdicts, with every other practice named. A veterinary practice adds the time the line ran and five fields of its own. Each is filled from the answer as written, then checked against your own pages.

line:             7 (rabbits, outside your hours)
ran:              [date] [time]
species for you:  [rabbits / "small animals" / none / wrong: ___]
service for you:  [as the answer words it / none]
hours stated:     [as the answer gives them / not stated]
  vs your pages:  [same / different: ___]
after hours:      [what the answer says happens after close / not stated]
  vs your pages:  [same / different: ___]
page cited:       [URL / a listing / none shown]

Species and service record what the answer attached to your name, in its own words. "Sees small animals" is not "sees rabbits", and on the rabbit line that difference is the reading. The two hours fields record what the answer told the owner and then the comparison, which keeps "not stated" apart from "wrong". An answer that states no hours gives the owner nothing to act on, which is a gap. One that states the wrong hours gives them something false, which is an error, and errors come first.

The page cited is the URL the answer links for the practice, its citation, when it shows one: your own page, a directory listing, a map profile. It turns a wrong hour into a task, because it points to where the hour came from. An answer that shows no source may have been written from what a model learned in training, so a blank in this field proves nothing about your pages.

Reading the hours fields before the score

Each month, read the hours fields across every line before you look at the score. Three patterns point to three different places.

• Different hours, with your own page cited. Your pages may disagree with each other (the footer says one thing, the contact page another), or the answer may reflect an older copy of the page. Check every page that states hours, then the Business Profile.
• Different hours, with a listing cited. The listing holds a copy of your facts that nobody updated. Correct it where it lives; when an answer gets your facts wrong covers tracing a fact back to the page that states it.
• After hours not stated, or stated as "open 24 hours" when you are not. Your pages do not say what happens after close in a sentence an answer can repeat, or say it only on a page nothing links to. That is the after-hours page, plus the same arrangement on the profile and in the markup.

The species and service fields come next, line by line. Where answers name you for "small animals" and never for rabbits, the page for that animal is the next thing to write; a page for each service you want asked about sets out what goes on it. The overview for veterinary practices lists the rest of what a practice's pages have to state.

What we can say about veterinary practices in general is little, so we say it plainly. In our audit of 9,483 small business websites, one buyer question per business went to Claude with web retrieval on, for 9,240 of them across 40 trades, and 14.9% of the businesses were named. It is directional (one question, one assistant, one day), and we publish no veterinary cut of it. For your own practice, the reading that counts is your panel, run the same way every month.