One location story, told the same way everywhere

The deliverable is not a list of recommendations. It is the finished thing: for every page, the exact field, the exact copy that goes in it, and the exact block of structured data that goes with it, in a form somebody can paste without interpreting anything.

That format exists because of how this work actually dies. A twenty-page audit describing what ought to change gets read once, agreed with, and never implemented, because turning advice into copy is the expensive part and it lands on whoever is least able to do it. Handing over the copy instead removes the step where the project stalls.

It also makes the work checkable. You can see exactly what will be on each page before it goes anywhere near the site, which matters when the person approving it is a director with a season to run rather than a marketer.

One canonical location story, enforced everywhere it appears: every title, every heading, the body pages, the map listing and its category, and the directory entries. The name spelled one way. The address written one way. The service area described in the same words in all of them.

This sounds administrative and it is the single highest-leverage thing available to a multi-location operator. A machine deciding which organization a query is about is looking for corroboration, and a brand that describes itself three different ways supplies three weak candidates instead of one strong one. Fixing it requires no new content at all — just a decision about which version is true, and the patience to make everything match it.

The structural mistake in this industry is collapsing four different things onto one page: the brand, the individual programs, the physical locations, and the age or skill groups. A parent does not search for your brand. They search for a sport, an age, and a place — and if all of that lives on one page, that page competes with itself for every one of those searches and wins none of them convincingly.

Each of those deserves a page that exists to answer one question. A program page that says what the program is, who it is for, what it costs, when it runs and where. A location page that is genuinely about that location rather than a template with the city swapped. The parent arriving from a search about a nine-year-old and a specific sport should land on the page about exactly that, not on a homepage that mentions it.

Demand in this industry arrives in waves that are entirely predictable and that almost nobody prepares for. Registration searches spike weeks before a season, camp searches spike well before the school holidays, and the sites that capture them are the ones whose pages already existed when the searching started.

Both indexing and citation lag publication. A page put up the week demand arrives has missed it. So the content calendar runs against the season rather than alongside it, and the boring version of this advice is the whole of it: publish the spring pages in winter.

The searches in this market are unusually literal. What time does it run. How old does my child have to be. Is there a beginner group. Where do I park. What happens if it rains. Do you offer a trial. How much is it.

Those belong on the page, written in the words people use, and matched by structured data that states the same thing rather than something adjacent to it. Schedules and program details are the two page types that get searched for most and that get neglected most, usually because they change often and nobody wants to maintain them in two places. Which is an argument for generating them from one source, not for leaving them off.

Those same questions now get typed at an assistant, which is a harder room. A results page gives a parent ten options and somewhere to scroll; an assistant gives two or three names and stops. Getting into that set is not a trick. It is structured data that states what the page is, a name and an address written identically everywhere they appear, and answers phrased so one sentence survives being lifted out of the paragraph around it. We report where you stand as a count rather than a position: three questions, three runs each, in each of ChatGPT, Perplexity and Google’s AI mode, and how many of those nine named you, on what date and which sources they cited. There is no position inside an answer to report.

Before any of it, a baseline. We run the AI-visibility protocol against the questions your market actually asks, so there is a real before to measure against later, and so that if an assistant is already naming a competitor in your city and not you, that is something we both know at the start rather than something that surfaces in a readout months in.

Audit first, then the package. The audit establishes which version of your location story is currently winning and where the disagreements are; the package turns that into the copy and the markup for every page. Sequenced that way because the audit frequently changes what the package should say, and because you should be able to stop after the first part if you do not like what you see.