Location Pages SEO: What 116 City Pages Taught Us About Local SEO at Scale

Every agency that sells local SEO for multiple locations eventually faces the same temptation. You have a service, you have a list of cities, and you have a page builder that can multiply one page by fifty. The maths is irresistible. So we did it, on our own site, and then measured what happened. This is that measurement, and the rules we now follow because of it.

It sits inside our wider work on how search changed this year. The GEO vs AEO vs SEO guide covers the landscape; our AI Overviews test found something about city queries that matters here too.

What Does Google Say About City Pages?

Two policies and one self-assessment question cover it, and none of them are vague.

Doorway abuse. Google’s spam policies define it as “when sites or pages are created to rank for specific, similar search queries,” and the first example given is “having multiple domain names or pages targeted at specific regions or cities that funnel users to one page.” That is a description of a city-page template with the city name swapped in and the same contact form at the bottom.

Scaled content abuse. Defined as “when many pages are generated for the primary purpose of manipulating search rankings and not helping users,” with the example of “using generative AI tools or other similar tools to generate many pages without adding value for users.” A page generator is one of those similar tools.

The helpful content question. Google’s own checklist asks: “Are you producing lots of content on many different topics in hopes that some of it might perform well in search results?” Swap “topics” for “cities” and it is the whole strategy.

None of this says location pages are banned. Google’s Business Profile rules assume you serve customers in many places. What is banned is the shortcut: pages that exist to catch a query, not to tell someone in Durham something about Durham.

What We Built and What We Measured

In mid 2026 we used a multi-page generator on digitalrhodium.com to publish two page sets: SEO services and PPC management, each for 58 US cities. One template per service, one CSV of cities, 116 URLs. Each page ran to about 1,800 words with the city and state substituted from the dataset. Diffed against each other, two pages in the same set were identical apart from the city token.

They indexed fine. Search Console showed impressions within weeks, mostly on PPC queries like “ppc management orem” and “baton rouge ppc management.” What they never showed was a click. Over 58 days, across all 116 pages, the click count was zero.

On 21 August we replaced 14 of them with individually written pages: six PPC cities and eight SEO cities, chosen because Search Console showed real demand there. Same URLs, same design system, same length. The difference was the content: a market introduction for that city, the industries that actually advertise there, what a realistic budget looks like in that auction, and FAQs a buyer in that city would ask. Measured by shared sentences, the Orem and Richmond PPC pages now overlap by about 26%. Before, they overlapped by 100%.

Then we watched two 20-day windows in Search Console: 1 to 20 August, while the 14 URLs were still template pages, and 21 August to 9 September, after the rewrite.

The rewrite roughly doubled impressions and moved the average five positions. The two pages with the most demand moved the most: Orem PPC went from position 53.6 to 43.7, Richmond PPC from 56.9 to 46.7 with six times the impressions. Not every page improved; Baton Rouge slipped four places and Durham held flat. And there is a bigger caveat we owe you.

What this does not prove. In the same weeks we also fixed the H1 on every city page, added Service schema, published four articles and started linking the city pages from the services page. We cannot isolate the rewrite from those. What we can say is that the 102 pages that got every one of those fixes except the rewrite stayed at position 75, and most of them stayed invisible. And that on a domain with an authority score of 2 out of 100, position 52 still produces no clicks. This is a direction, not a destination.

Why Do Identical City Pages Fail?

Three reasons, and only one of them is about penalties.

There is nothing to rank. Modern retrieval works at the passage level, a point we made in the GEO guide and that applies to classic search too. For “ppc management durham,” Google needs a passage about PPC in Durham. A template page has a passage about PPC that mentions Durham once in the heading. Every competitor who wrote even two real paragraphs about the Research Triangle has a better passage.

Duplicate pages compete with each other and with your real pages. Our generic PPC services page picked up 360 impressions for city queries at position 88, because Google could not tell which of the 59 near-identical pages was the answer. Once we gave the six real PPC pages a “by city” block on the services page, with exact-match anchors, that confusion had somewhere to resolve.

Thin pages set the site’s quality baseline. A site that is 116 template pages and 4 articles is, to Google, mostly template pages. Google’s helpful content guidance is explicit that site-wide signals exist. Every real article we publish is diluted by the 102 pages that are not.

A test we now apply before any page goes live: what would you write for this city that you would not write for the last one? If the honest answer is “the name,” the page should not exist yet.

Can a Service-Area Business Rank in Every City’s Map Pack?

No, and it is worth being precise about why, because a lot of multi-location SEO advice quietly assumes you have an office in every city.

Google’s Business Profile guidelines say service-area businesses “should have one profile for the central office or location with a designated service area,” that you “should hide your business address from customers,” and that the service area “shouldn’t extend farther than about 2 hours of driving time from where your business is based.” A rented mailing address does not help: “a virtual office” is “not eligible for a Business Profile,” and a co-working desk only counts if it has signage, receives customers and is staffed by your people during business hours.

For us that means one profile, in Houston. For the other 57 cities, the map pack does not exist as an option, and any agency promising it is promising a profile Google will suspend.

That reframes what a remote city page is for. It is not supporting a map listing. It is the entire presence. And our AI Overviews test found that on city-plus-service queries Google shows the map pack and skips the AI summary, so the organic results under the pack are where a remote business lives. A template page competing there against local agencies with real profiles is bringing nothing to the fight.

What Makes a Location Page Real?

Here is the checklist our 14 pages were rebuilt against. It is the practical version of “would you write this for another city.”

1. A market paragraph that could only be about this city. For Baton Rouge PPC, that meant the petrochemical corridor, the legal auction, and an LSU calendar that moves consumer demand every semester. For Orem, it meant bidding against funded software companies twenty minutes up the road in Lehi.
2. Industries chosen for the city, not pasted from the last one. Providence gets marine and coastal trades. Durham gets life sciences services. If the list is the same eight industries everywhere, it is decoration.
3. Numbers that differ by city. Budget ranges, auction competitiveness, where the demand concentrates. Our ChatGPT citation test found specific numbers were what got small sites cited; the same detail is what makes a city page credible to a person.
4. FAQs a buyer in that city would actually ask. “Should a Baton Rouge industrial firm run search or social” is a real question. “What is PPC” repeated 58 times is not.
5. Honest schema. Service markup with areaServed naming the city, provider pointing to your organization. Not LocalBusiness with an address you do not have; our schema guide explains why that ends in a suspension.
6. A path in and a path out. The services page links to each real city page by name. The city page links back to the service and to the relevant guides. A city page nobody links to is an orphan whatever its quality.
7. A uniqueness floor you can measure. We diff each new page against its nearest sibling and want at most a third of sentences shared. Shared boilerplate (the process, the CTA) is fine; shared substance is not.

Two things we deliberately did not do. We did not invent local statistics; a made-up figure is worse than none, and we removed one that had crept into the old template. And we did not use client results, because our clients did not agree to be marketing material.

How Many Location Pages Should a Website Have?

As many as you can write to the standard above, and no more. That number is usually far smaller than the number of cities you would like to rank in.

Our own order of operations, which we would recommend to any service-area business:

1. Start from demand, not from a map. Search Console told us which cities were already producing impressions. Orem, Baton Rouge, Richmond, Durham, Providence and Raleigh were the PPC cities with real queries; that is why they were first.
2. Write the city you are physically in properly, first. It is the only one where a page and a profile can work together. Ours is Houston.
3. Rewrite in batches of six to eight, then wait a month and measure before the next batch. It took us about a working day per page to do it honestly.
4. Decide what to do with the remainder. The candid options are to noindex template pages until they are rewritten, or to leave them indexed and accept the drag on the site while you work through them. We are currently doing the second and re-evaluating monthly; the impressions they earn are close to worthless, but they map where demand exists.

When Does Programmatic Local SEO Actually Work?

It works when the data behind each page is genuinely different per city, and the page is a presentation of that data rather than a pretext for a keyword. Property listings, weather, event calendars, store inventory, pricing that varies by market, regulations that differ by state. In those cases the template is honest: the reader in Durham really is getting different information from the reader in Orem, because the underlying record is different.

Service businesses rarely have that data. What they have is the same service delivered in different places, and the only thing that varies is the market context, which has to be written by a person who has looked at that market. That is why our template failed and our rewrite did not: the generator was being asked to invent differences that did not exist in the dataset.

The honest use of a page generator for a service business is scaffolding: it builds the URL, the layout, the schema and the navigation, and a person supplies the paragraphs that make the page about somewhere. That is how our 14 pages are built now, and how the next batch will be.

Frequently Asked Questions

← Back to all articles