Why your city pages stopped ranking in 2026

The March 2026 core update completed on April 8. It had the largest impact on local service businesses of any update this year, and within that, sites built on templated location pages took the worst of it.

Home services, legal and healthcare saw the biggest ranking shifts of any verticals. If you are a roofer, an HVAC company or a plumber who watched organic traffic fall through April and never quite recover, you are in the group this hit.

We have been hired three times since then to clean up exactly this. Here is what we find, and what we do.

What a templated city page looks like

You probably have them. An agency sold a package described as “local SEO coverage” and delivered a set of URLs like this:

• /roof-replacement-katy-tx
• /roof-replacement-cypress-tx
• /roof-replacement-tomball-tx

Open two of them. The headline changes. The city name appears eight times instead of the other city name eight times. A stock photo of a roof appears on both. The body copy is the same four paragraphs about your commitment to quality and your free estimates.

That is the pattern. Google has been able to detect it for the better part of a decade and has told people not to do it for nearly as long. What changed in March 2026 is that the cost went from a slow leak to a cliff.

Why it stopped working

Three things happened at once in that update, and city pages sat at the intersection of all three.

Templated content got re-evaluated harder. The update reinforced a push toward entity-based local authority over keyword matching. A page that exists only to contain a keyword and a place name has nothing to offer that system. It is not authoritative about the place, because it says nothing about the place.

Thin pages started dragging on the site. A large set of near-identical pages does not fail in isolation. It affects how the site as a whole is assessed, which is why businesses in this position often find that their strong pages also slipped, not just the templated ones. That is the part that surprises people. You did not lose eighty pages. You lost the site’s standing.

Crawl budget got spent badly. Eighty thin pages compete with your genuinely good pages for attention. This matters less than the first two, but it is real, and it is why consolidation often produces faster improvement than anyone expects.

What we actually do about it

Not what most people expect. The instinct is either to delete everything or to rewrite all eighty pages. Both are usually wrong.

Step one: find out what is load-bearing

Before touching anything, pull every location URL with its impressions, clicks, rankings and conversions for the last twelve months. In every cleanup we have run, the distribution is brutally lopsided. Out of eighty pages, four or five are doing essentially all of the work and the rest have never produced a call.

You need that list first, because the pages that are working are usually not the ones the client assumed.

Step two: decide which markets you can actually speak to

For each market, ask one question: can we say something true and specific about working here that we could not say about anywhere else?

Real answers look like: we run six crews out of a yard twenty minutes from here, we did 140 roofs in this county after the May hail, this municipality requires a specific permit that adds four days, the housing stock here is mostly 1990s build with a particular tile that fails a particular way.

If you cannot answer, that market does not get a page. It gets a line on your service area and it gets covered by the profile work instead. This is the part clients resist, and it is the part that works.

Step three: consolidate, do not delete

Take the markets that survived step two and build genuinely specific pages for them. Usually somewhere between five and fifteen, not eighty.

Everything else gets a 301 redirect to the most relevant surviving page, which is normally the parent service page. Redirect, do not delete, because some of those thin pages have picked up links or have residual rankings worth keeping, and a 404 throws that away.

In one cleanup this year we removed more than thirty pages for a roofing client and consolidated the remainder into eight. Organic traffic recovered past its pre-April level within about ten weeks. That is one case and it is not a promise, because how fast this recovers depends heavily on what else is wrong with the site. But the direction has been consistent everywhere we have done it.

Step four: make the survivors worth ranking

A page that survives the cut still has to earn its position. What we put on them:

• Real scope. What the job actually involves in that market, including the parts customers do not expect.
• Real price context. Ranges, what moves them, and why. Contractors hate this. It converts better than anything else on the page.
• Real photos of real work in that market, where they exist. Stock imagery is a wasted slot.
• The local specifics from step two, stated plainly rather than padded.
• Internal links connecting service to market and back, so the relationship is stated rather than left for Google to infer.

And then the part most SEO work skips entirely: the page has to produce a phone call. Ranking a page that converts at one percent is a way to lose money slightly more efficiently. Where the proof sits, where the number sits, and what the page asks for are part of the SEO work, not a separate project for later.

The honest version of the trade-off

Fewer pages feels like less coverage, and contractors reasonably worry that a competitor with eighty pages will out-rank them.

Two things about that. First, after March 2026 that competitor is probably not out-ranking anyone. Second, and more importantly, the map pack is where the volume is for home services, and the map pack does not care how many city pages you have. It cares about your profile, your proximity and your reviews. Your site’s job in that system is to corroborate what your profile claims, and a thin templated set corroborates nothing.

So the trade is not coverage for no coverage. It is fake coverage for real coverage, plus a site that stops dragging on everything else.

How to check whether this is your problem

Ten minutes, no tools beyond Search Console.

1. Pull your location pages and sort by clicks over the last twelve months. If more than about 80% of them have produced fewer than ten clicks, you have a templated set.
2. Open two of them side by side. If you can read both and struggle to name a difference beyond the place name, so can Google.
3. Check the date your organic traffic turned. If it bends in March or early April 2026, the timing lines up with the core update rollout.
4. Check whether your map pack position moved in the same window. If it did not, the problem is site-side, which is consistent with what this update did.

If all four line up, you have your diagnosis, and the work above is the work.

What not to do

Do not rewrite eighty pages with an AI tool and republish them. You will have produced eighty pages of slightly different thin content, which is the same problem with a larger word count. The update went after pages that offer nothing unique, and volume is not uniqueness.

Do not delete everything at once and hope. Redirects exist for a reason.

And do not add another forty cities because coverage feels like the answer. Whatever is already true about your site gets multiplied by scale, including the problems.