How many location pages is too many?

There is no fixed number - the limit is how many markets you can supply genuine facts for. A regional contractor with real crews and real job history across forty towns can justify forty pages. A single-van operator cannot justify eight, regardless of how far they will drive. The count is a symptom; the facts are the thing being measured.

Can I use AI to write location pages?

For drafting and structuring, yes, and we do. What it cannot do is invent the five fields, because those are facts about your business that exist only in your dispatch system and your crew's heads. A model given no local facts produces exactly the fluent, structurally identical prose that gets these pages filtered, so the intake call is the part that cannot be automated away.

What if I genuinely serve forty towns?

Then either do forty intake conversations over time and publish as the facts arrive, or publish a smaller set of genuinely strong regional pages that name all forty honestly. Publishing forty thin pages because you serve forty towns is the failure mode - the service area is real, the individual pages are not.

Do location pages still work at all?

Yes, and well, when they carry something real. The pattern being filtered is not the concept of a page per market, it is a page per market with nothing on it. A location page that names a job, a technician, a drive time and a local condition is a genuinely useful document, and those continue to rank through updates that flatten their thin equivalents.

Should I delete thin location pages I already have?

Consolidate rather than delete outright, because deletion throws away whatever links and history the URLs accumulated. Redirect the thin ones into a regional page that covers them properly, so the equity moves somewhere useful and the pattern stops being visible as a set.

Will a service-area business rank without a street address?

In the map pack it needs a verified profile, which a service-area business can have with a hidden address - that is the intended configuration and it works. Organic location pages have never required an address at all. What they require is content that could only have been written about that place, which is the same rule as everything else here.

How long does one location page take to produce properly?

About ninety minutes of the client's time per territory for the intake, plus the writing. That is the real constraint and the reason the output is measured in a handful of pages a month rather than dozens. It is also considerably cheaper than rebuilding a filtered site, which is the alternative being priced against.

Does internal linking rescue a thin location page?

No. Links help a page that deserves to rank get found and get weight; they do not supply the substance that is missing. A well-linked page with nothing on it is a well-linked page with nothing on it, and adding links between a set of near-identical pages tends to make the set more obvious rather than less.

Location Pages That Survive a Spam Update

The short version

The location pages that get wiped out by spam updates are all built the same way: one template, a spreadsheet of place names, and a find-and-replace. That pattern has been detectable for a decade, and the sites carrying hundreds of such pages go down together rather than one at a time. The fix is not better writing on the same template - it is a rule that a page cannot publish unless it carries facts that exist nowhere else on the site. We require five, enforced at build time, and the important consequence is that the rule caps how many pages you can produce. That cap is the feature. A business that can honestly fill five fields for eight towns should have eight location pages, not sixty.

Why templated location pages get filtered

Every mass-filtered location page network looks identical under the hood. One layout, a list of place names, and prose where the only thing that changes between pages is the noun. It is trivially detectable - the pages share sentence structure, paragraph length, heading order and often word count to within a few percent - and the detection has existed far longer than any recent update.

What makes it dangerous is that the failure is correlated. These pages do not degrade individually; they are recognised as a set and demoted as a set, which is why the experience of a spam update is not 'a few pages slipped' but 'four hundred pages stopped ranking on the same afternoon'. A site that leans on them for most of its traffic loses most of its traffic at once.

The five fields every page must have

So we require five unique fields per page before it is allowed to publish, and none of them can be generated from the place name.

A local job reference with a real address range, because it proves the crew has been there. A neighbourhood-specific problem note - hard water, hail exposure, panel age in that build era, clay soil movement - because it demonstrates knowledge that a template cannot fake. The actual technician who covers that area, because a named human is a fact. A service-area polygon rendered as a map, drawn on real drive times rather than a radius. And the average dispatch time from the closest depot, which is the single most useful number a person deciding between two contractors can have.

Why it is a build gate rather than a guideline

The distinction that matters is enforcement. Every agency says its location pages are unique; almost none can tell you what happens when one is not. Ours is a validation function that runs at build time and throws - if any of the five fields is missing, or if the prose duplicates another entry after normalising whitespace and punctuation, the site does not build.

That is deliberately annoying. A guideline degrades the moment somebody is in a hurry at page thirty-one, and page thirty-one is exactly where the copy-paste starts. A build error does not degrade, and the person in a hurry is forced to either find the fact or drop the page.

The cap is the point

The predictable objection is that the rule limits output, and it does. Filling five genuine fields takes roughly a ninety-minute intake call per territory, which is why we would rather ship eight good location pages a month than sixty empty ones. Most agencies sell the sixty because sixty sounds like more work delivered.

But the cap is doing something valuable: it is telling you the truth about how many markets you actually serve. A business that cannot name a job, a technician and a dispatch time for a town does not really serve that town - it would drive there if asked. Those are different things, and a location page is a claim about the first.

What to do with markets that fail the test

Do not publish, and do not publish a thinner version. The instinct is to write something shorter for the marginal towns, which produces exactly the pattern the filter is built to catch, only with fewer words. If a market matters enough to want a page, it matters enough to do one intake call for; if it does not, it belongs in a service-area sentence on a page that does exist.

The other option worth considering is a genuine regional page - one page covering the corridor, the county or the cluster, carrying real facts about all of them - rather than five thin pages pretending each town is a separate market. That page can be excellent. Five pretending cannot.

How to tell whether yours are already thin

There is a test that takes ten minutes and is uncomfortably decisive. Open three of your location pages, copy the body text into a document, and delete every instance of the place name. Then read them. If you cannot tell which page is which, neither can a ranking system, and no amount of internal linking or schema will change that - the pages are the same page with a different label.

The second check is Search Console's index coverage. Pages that are crawled but not indexed, in volume, on a location template, is the clearest signal available that the pattern has already been recognised. That state is not a penalty and does not need a reconsideration request; it is simply a judgement that the pages are not worth storing.

What we do instead when a client wants coverage everywhere

The request behind sixty location pages is usually reasonable. Somebody wants to be findable in every town they will drive to, and pages felt like the only way to say so.

There is a better way to say it. One honest service area page, with a real map of where the crews go and the drive times from each depot, covers the whole catchment without pretending each town is a separate operation. It answers the question a person actually has, which is whether you will come out here, and it does it in one place you can keep accurate.

Then build proper pages only for the markets where you genuinely compete, and add one each time a new market earns it. That is slower on paper and faster in practice, because the pages that exist all work, and none of them are dragging the others down by association.

Check it on your own site

Run the place-name deletion test on three pages
Copy three location pages into a document, delete every occurrence of the town name, and read them side by side. If they are indistinguishable, you have one page published under several addresses, and that is the whole diagnosis.
Count the facts, not the words
On a single page, count the assertions that could only be true of that market - a street, a technician, a drive time, a local condition. Word count is irrelevant; a four-hundred-word page with five real facts outperforms a twelve-hundred-word page with none.
Check crawled-but-not-indexed in Search Console
Filter index coverage to the location template. A large block sitting in crawled-currently-not-indexed means the pattern has already been assessed and found not worth storing, which is information you can act on rather than a penalty to appeal.
List the markets you can genuinely fill five fields for
Write the list before writing any pages. It is almost always shorter than the service-area list, and the difference between the two lists is the set of pages that were always going to be padding.
Decide what happens to the ones that fail
Consolidate them into a regional page carrying real facts about the cluster, or fold them into a service-area sentence elsewhere. Leaving them published while adding good pages alongside means the good ones keep the company of the pattern that is holding them back.

Programmatic location pages that survive a spam update