Why discoverability is a record problem, not a ranking problem
Most clinics that worry about being found treat the question as a contest. They ask who is above them, what those clinics are doing, and how much it would cost to overtake them. That framing produces anxiety and very little action, because the ranking of a search result is an output of a system nobody outside it can see.
There is a more tractable question underneath it. Before anything can rank, a search engine has to hold a description of the clinic, and that description has to be right. The address has to be the address. The hours have to be the hours. The services listed have to be services the clinic performs. When those facts disagree across sources, the engine has to choose which to believe, and it will sometimes choose the wrong one. That is not a competitive problem. It is a record-keeping problem, and record-keeping is something a clinic can fix in an afternoon.
This instrument therefore assesses the record. It asks whether each published fact about the clinic is correct, whether it is the same everywhere, and whether somebody at the clinic is responsible for it. Those three properties are within your control regardless of what any other clinic does. A clinic that scores well here has not beaten anybody. It has simply stopped losing patients to details it never checked.
The distinction matters for a second reason. Ranking advice ages badly. Guidance on how a particular result page is assembled is out of date within months. Record accuracy does not age. A correct address was correct in 2015 and will be correct in 2030, and every system that ingests it, including systems that do not exist yet, benefits from the same discipline.
How to score this instrument
Work through the twelve criteria in order, scoring each 0 to 3 against the descriptor written beside it. The descriptors are written so that 0 and 3 are unambiguous. Where you genuinely sit between them, score 1 if the thing exists but is neither documented nor owned, and 2 if it is documented or owned but not both.
Score what is true today, not what is planned. A profile that somebody intends to claim scores 0. A directory sweep that is on next quarter's list scores 0. This is uncomfortable and it is the point: an assessment that credits intentions produces a number that flatters the clinic and helps nobody.
Score with the evidence open in front of you. Open the profile, open the website, open the directories, and check each claim. Scoring from memory reliably produces a total four to six points higher than scoring from evidence, because people remember the state of things at the moment they last thought about them.
Two people should score independently where possible, then reconcile. Disagreement is informative. If the practice manager scores hours accuracy 3 and the receptionist scores it 0, you have found something more useful than either number.
Record the total and the date. The value of this instrument is almost entirely in the second run: a total of 22 means little on its own, and a great deal when the previous run was 14.
Common scoring errors
Scoring the intention rather than the artefact. The most frequent error by a wide margin. "We know our address is consistent" is not evidence. Open the listings.
Treating the website as the record. The website is one publication of the record. Patients arrive through profiles, maps, aggregators and directories that were populated years ago from sources nobody remembers.
Scoring service coverage generously. A profile that lists a treatment the clinic stopped offering is worse than a profile that omits it, because it generates enquiries the clinic must decline. Score 0 if anything listed is not currently provided.
Confusing volume with accuracy. Appearing in forty directories is not a score of 3 on duplicate listings. It is usually a score of 0, because nobody has checked what those forty entries say.
Scoring structured data by presence rather than validity. Markup that is present but contains a different phone number from the profile is scoring against you, not for you. Validate it, and check that the values match the source of record. The local business structured data documentation sets out what the fields expect.
What to do after scoring, by criterion rather than by band
Bands describe a state. Actions belong to criteria. Once you have a total, ignore it for a moment and look at which individual rows scored 0 or 1, because those are the work.
If profile ownership scored low, everything else is provisional. A record you cannot edit is a record somebody else controls. Recover access before touching anything downstream.
If name, address or contact consistency scored low, write the source of record document first. It takes twenty minutes and it converts an open-ended tidying exercise into a checklist, because every listing is then either identical to the document or it is not.
If hours or service coverage scored low, the fix is a cycle rather than an edit. Assign the review to a named role with a fixed frequency, and record the date of the last check somewhere visible.
If structured data scored low, treat it as a website task and pair it with the technical foundations audit, which covers the same territory more thoroughly. If duplicates scored low, budget a half day: finding legacy listings is slow, unglamorous and worth doing once properly.
When the low rows are closed, move outward. The intent coverage gap analysis asks a different question: not whether your record is right, but whether you have anything to say to the people the record brings you.