Anaks Labs
Article 03

When it names somebody else

You asked an assistant for a dentist in your own town and it recommended the practice four minutes down the road, warmly, at length, with their opening hours attached. This article is about what that observation is worth. It is worth a good deal — but not most of the things people immediately conclude from it.

Published · Anaks Labs

What it actually looks like

It rarely arrives as a single insult. It arrives as a shape you notice after the third or fourth attempt, and the shape usually falls into one of three kinds.

Sometimes it is one practice, every time, whatever you ask. Sometimes it is a rotating cast of four or five in the area, none of which is yours. And sometimes it is a chain or a hospital group, because a large organisation publishes a great deal of text about itself and there is more of it to draw on. The second kind is the most common and it is also the one people notice least, because no individual reply looks like an omission until you line six of them up.

Whichever kind it is, the observation to hold on to is not "they were praised". It is "somebody was named and it was not us", which is a different and more useful finding.

Six things this is not

This section is the point of the article. Practice owners lose money on the wrong conclusion here far more often than they lose it on the underlying problem.

  • It is not a penalty

    Nothing has been applied to you. There is no strike against your practice, no flag, no blacklist, and nobody has removed you from anything. Being absent from an answer is not the same event as being demoted from one, even though it feels identical from where you are standing.

  • It is not a verdict on your dentistry

    The quality of the work you do is not among the things an assistant is in a position to assess. It has never seen a patient of yours. Whatever it said about the other practice, it said on the basis of text, and no amount of clinical excellence is text until somebody writes it down somewhere reachable.

  • It is not a reflection of your reviews

    Owners assume the practice being named must have better ratings. Frequently it does not, and you can check that in about a minute by looking both of you up. When the star ratings are close, or the other way round entirely, the review theory is finished and something else is going on.

  • It is not something they bought

    There is no placement to purchase inside an assistant's answer in the way there is a sponsored slot above a search result. If a vendor tells you your competitor is paying for that position and that you can too, you have learned something important about the vendor.

  • It is not stable, and it is not a ranking

    Ask again next week and the cast may change. There is no position to occupy, no number to be at, and consequently nothing to report on month over month in the way an agency reports keyword positions. Anyone selling you a rank inside an AI answer is selling a metric that does not exist.

  • It is not necessarily about anything you can see on their website

    Open the competitor's site and it will very often look worse than yours. That is not a paradox and it is not evidence that the whole business is arbitrary. It means the property being compared is not the property you are looking at, which is the subject of a separate article.

The finding that sits closest to it

Our field work covered 182 clinic websites in the United States, and we read what each of them hands over. Across every group we could tell apart, at worst two in three of them never answered a single question a patient would type — cost, insurance, recovery, availability, parking — anywhere on the page. The sites that had already been singled out by other people as well-made examples were inside the same band as the rest.

Anaks Labs field work, 182 US clinic websites in two separately drawn groups. We publish this as a floor rather than a single figure, and the reason is in the scope note. Sample, definition and limits → · how the two groups were drawn →

A page that answers no question cannot be the source of an answer. That is as far as this article goes, and the stop is deliberate: the sentence after that one is a set of instructions, and the instructions are the thing we are paid for. We would rather tell you plainly where an article ends than publish a half-recipe that leaves you worse off than before you read it.

What to write down

If you are going to do anything with this observation, it should be to turn it into a record. Four things are worth capturing, and all four are free.

  • Which practices get named, and how consistently

    One name recurring is a different finding from five names rotating. Note the split.

  • Which questions they get named for

    Whether the other practice appears for everything, or only for particular questions — implants, emergencies, children, a specific insurer. A narrow pattern is a much more informative one.

  • Whether your own site is ever cited, for anything

    Where the assistant shows what it drew on, note whether your domain appears at all, even in an answer that names you. Never appearing is the strongest single signal in the exercise.

  • The date, and what you asked, word for word

    Because the moment anybody touches your website this becomes a before, and a before is only useful if it was written down while it was still true.

Then leave it alone. Asking an assistant about your practice repeatedly does not improve your standing, and no subscription that monitors your "AI rank" is measuring a thing that exists. How to run the test properly, and how to avoid over-reading it →

Read next

The audit this article draws on, with both sample groups and what each one was drawn from →