The biggest local build in my case studies is a locksmith network with more than 90 town and service pages, designed, built, and pushed into Google from the same desk. People ask why 90 and not 20, or not 500. It is not a marketing number. It is a map.
A service business does not serve “the city.” It serves forty towns inside that metro, plus the suburbs nobody thinks to name, plus the county-line places every competitor ignores. Draw the real radius the vans actually drive, put a pin in every town inside it, and the page count falls out on its own. For that locksmith cluster the map said 90 plus. For a California mobile notary covering counties and regions, it said 250.
The math that actually matters
Towns are only the first multiplier. Multiply by the services each town orders, and the set grows. Multiply by languages and it grows again: a dietitian practice running English, French, and Ukrainian versions came to 220 tracked URLs, because every language version of every page has to earn its own place in the index. All 220 are indexed at the last snapshot, and none of them got there by being uploaded and forgotten.
That is the part the page count hides. Ninety pages is not ninety pieces of writing. It is ninety URLs that each need to be discovered, crawled, judged, and stored, and the judging is where templated town pages die. Google’s quality systems were built to catch exactly this shape: many URLs on one template where only the place name changes. Publish them carelessly and they pile up in Crawled, currently not indexed, which is Google’s polite way of saying it has seen this and is not impressed.
Waves beat bulk, every time
The rollout schedule matters as much as the writing. A few hundred near-identical URLs appearing overnight, linked from nowhere, is the exact signature of a doorway-page farm, and Google samples it accordingly. A wave of fifteen or twenty genuinely local pages, published, linked, submitted, and proven indexed before the next wave ships, reads like a real business expanding.
Publishing at random does not work. Publishing in waves, with real local proof on every page, does. Each wave gets its own crawl, its own check, its own chance to build the site’s standing so the next wave lands easier. When a wave underperforms, that is the moment to stop and diagnose, not to ship the next one on schedule.
What separates the pages that index from the ones that stall comes down to three things, and they are the same three on every job:
- Genuine local proof. The areas actually covered, response times from the real base, the services that town actually orders. The test from the location pages guide: cover the town name, and you should still be able to tell which town the page is about.
- Internal links that follow geography. Neighboring towns link to each other, service pages link to towns, towns link back. Not ninety orphans hanging off one hub.
- A repeating structure without repeating substance. The skeleton can be shared. The sentences cannot.
What the ledgers actually show
Every one of these jobs ends in a per-URL ledger, so here is what the last snapshots look like across five local and service-site builds:
The plumbing service sits at 110 of 110. The medical practice closed at 56 of 56, tracked one URL at a time. The college dorm store, a mixed content and commerce site, reached 295 of 295 with new posts submitted as they were published. The fitness studio in Australia holds 71 of 73 with the class and training pages, where sign-ups actually come from, prioritized first.
And the locksmith flagship, the biggest and youngest set, stands at 138 of 193 and climbing, with one of the smaller sites in the same cluster at 41 of 41. I publish that 72 percent as readily as the full hundreds, because a ledger you can open beats a rounded-up percentage in a slide deck.
The 3 percent is the point
The notary job is the cleanest lesson in why the ledger matters. 250 pages, 243 indexed, 97 percent. The interesting part is not the 97. It is that the remaining seven pages are known by URL. Finding a handful of county pages sitting in Crawled, currently not indexed is not a disaster. It is a diagnosis, and it is a far better position than a site owner who only knows an average.
That is the standard worth holding anyone to, my own work included: a URL-by-URL record with dates and statuses, like the 435 of 435 tracked for a health review publisher. Whoever builds your town pages, ask to see that spreadsheet. If they cannot hand it over, they are selling adjectives.
Where this work comes from
The locksmith cluster is also the reason the builds page and the case studies overlap: pages, design system, and indexing came from one desk, which is work I do end to end. If your town pages already exist and are stuck, that is the other half of the desk: the indexing service starts with naming which of the three failure patterns above is yours, free, usually within a day. Bring the map, or bring the stuck pages, either works.