Yes — if you want both “furnace repair in City A” and “furnace repair in City B” to rank, they need to be two different pages, and each page needs genuinely unique local content: real neighborhoods, real jobs, real reviews from that area. A location template with the city name swapped is the documented failure mode — it produces a set of near-identical pages that read as thin scale, and it gives neither the search engine nor a customer a reason to prefer one page over another.

We get this question from local business owners and from the marketers who serve them, usually in the form: “We serve a 40-mile radius. Do we need a page for every town?” This post answers the structural question, names the failure mode practitioners actually hit, makes “unique local content” concrete with a worked skeleton, and explains why this is now a question about AI answers, not just Google. It’s part of the local-SEO series — the overview is what local SEO actually works in 2026.

Why one page, per city and per service

Search engines map pages to queries, not the other way around. A single “Cities We Serve” page that lists five cities is one URL competing for five “service in [city]” queries, and it can fully own at most one of them — usually the strongest. The structural fix is to make the URL match the query: one page per city and per service. A heating company serving three cities with three trades builds nine pages, each targeting one query, each standing on its own as an answer to “furnace repair in [city],” “AC repair in [city],” and so on.

The URL pattern usually looks like /services/furnace-repair/peoria/. The exact structure matters less than the principle: if a customer would phrase it as a question, the site should have a page shaped like the answer. Which of those city+service queries are worth a page at all — and which ones actually convert — is the subject of the companion post on bottom-funnel local keywords.

The failure mode: 50 near-identical location pages

The most common failure is not building one page per city. It’s building fifty of them from one template. When a business expands coverage, the fastest path is a generator that emits a page per city with the city name swapped into the same sentences: “We provide [service] in [City]. We have served [City] families for over X years. Contact us today.”

Local-SEO practitioners report hitting this over and over: coverage expands, pages multiply, and the page set goes to pieces much faster than expected — same structure, same claims, same body copy, one variable. The honest way to label it: fifty of those pages does not make the business look like it serves fifty cities. It makes the site look like fifty copies of one page. Google’s spam policies name “creating substantially similar pages” as doorway abuse, and in practitioner reports the usual outcome for a set of pages that differ only by a swapped name is that most of the set stays invisible — sometimes with the whole set demoted.

We present this as practitioner observation, not a controlled study: the consistent report is that the templated pages never earn the results the single strong page did. The fix is never “add more of the same page.”

What “unique local content” actually means

Unique means different facts, not different words. Rewriting the template with a thesaurus is still the same page. Four kinds of content make a city page genuinely different from its siblings, and each one has to be true before it goes on the page:

ElementWhat it looks like on the pageWhy it matters
Neighborhoods, with real detailNamed neighborhoods, each with one or two specifics — housing stock, travel time, a common local problemMatches how locals actually search (“furnace repair in [neighborhood]”) and proves the business knows the geography
Local landmarksNamed reference points customers recognize — a riverfront, an arena, a district, a highway entranceSpecifics a template cannot produce; they corroborate that the business is actually there
Specific local jobsTwo or three recent jobs, each naming the neighborhood, the problem, and what was doneFirst-hand evidence a competitor’s template page can’t copy; also the content AI answers can quote
Local reviews, tied to the areaReal customer quotes, tied to a neighborhood wherever possibleFeeds and is fed by the review-velocity system for local businesses; replies to those reviews are local content too

The test is simple: read the page and ask what a customer in a different city would find here that they couldn’t find on any other page in the set. If the answer is “the name,” the page isn’t unique.

Worked example: a city page skeleton

The skeleton below is an example, clearly not real content from any real business. City, neighborhoods, jobs and reviews are illustrative — the point is the granularity each section demands, not the facts. It is for “furnace repair in Peoria, IL,” a mid-size city, so the pattern transfers to any city of similar scale.

URL: /services/furnace-repair/peoria/ H1: Furnace repair in Peoria, IL

Lead (2–3 sentences). Who, where, and one specific: “Hartwell Heating repairs furnaces across Peoria and East Peoria, usually same-day. If your heat is out before a cold snap, we answer a phone, not a ticket system.” No “we are a family-owned company passionate about comfort” filler.

## Neighborhoods we cover. Four to six named areas, each getting one or two sentences of real local detail:

  • River Park — mostly homes from the 1950s–70s; older cast-iron furnaces still running, so heat-exchanger failures are the common call.
  • East Peoria — a short trip over the bridge; mixed stock, a lot of 20+ year furnaces that need annual tune-ups before winter.
  • North Peoria — newer builds with high-efficiency units and two-stage thermostats; calls skew toward controls and sensors.

Notice what each line is doing: it is a fact about that neighborhood’s housing that a furnace technician would actually know, and it happens to match a real query pattern (“furnace repair River Park”).

## Recent jobs in Peoria. Three entries, each one or two sentences: neighborhood, problem, fix, date. “East Peoria, October: 28-year-old furnace tripped the limit switch mid-week; heat exchanger was sound, sensor replaced, system back on the same day.”

## What Peoria customers say. Two or three real reviews, quoted with the neighborhood where possible. These come from the review system — asking after every completed job with a direct link — not from an inventory of stock testimonials.

## FAQ. Two or three questions that actually recur for this area: “Do you cover East Peoria?” “What does an emergency furnace call cost in Peoria?” “What brand do you stock?” Answers in one or two sentences each.

A page at this granularity usually lands around 600–900 words. Length is not the goal — the goal is that every section could not be copy-pasted, unchanged except for the name, onto a page for another city. If a section can be, cut it or find the real detail.

The AI-content angle: why templated pages get pruned

Pages that differ only by a swapped city name are exactly the pattern that gets pruned. Search engines have handled “scaled thin content” for years, and the 2026 position in the local niche is consistent: a set of near-identical pages is a remediation problem (audit, prune, merge, redirect, add first-hand detail), not a growth asset. We treat that evidence as observational, not causal — but the failure mode is well documented in practitioner accounts, and it is not hard to recognize in a client’s site.

The same logic now runs on the answer side. When a customer asks an AI surface “who fixes furnaces in Peoria,” the system grounds its answer in pages it can actually cite — extractable, specific, and checkable: named neighborhoods, dated jobs, reviews tied to a place. A template page offers none of that, so it is not even a candidate. The mechanics — how answers retrieve, select, and cite — are laid out in our AEO and GEO one-page primer; the local version of “citable” is simply “specific and verifiable about that place.”

This is why the two questions — will these pages rank in Google and will these pages ever be cited by an AI answer — now have the same answer, and the same requirements: one page per city per service, real unique local detail on each.

Keep shared info structured, not copy-pasted

The DRY rule: hours, phone number, address, service area, and any other info that belongs to the business rather than the city appears exactly once and is rendered everywhere.

  • In the body, shared info is absent. Do not paste the hours block into the bottom of forty city pages. That is shared content in the wrong place, and it dilutes the unique signal each page carries.
  • In the structure, it lives once. Site-wide footer, one LocalBusiness (or Service) JSON-LD block per page that points at the same canonical facts, and one source of truth — a settings page or CMS field — that you edit once when the phone number changes.
  • In the schema, each city page adds its local specifics: the city in the areaServed or Service markup, not a restatement of the boilerplate.

This matters for AI answers as much as for Google: recombined facts come from consistent structured sources, and a business whose hours appear in four slightly different phrasings across its own pages is the business that gets quoted wrong.

If this is your situation

If you have a location-page set and you’re not sure whether it is a set of pages or one page fifty times, the honest audit takes an afternoon: open the ten weakest pages side by side, and count how many sentences survive a find-and-replace of the city name. If the count is high, the fix is the prune-and-rewrite path — merge the weak ones, redirect them, and rebuild the survivors around real local detail — not more pages and more traffic. We run that audit as part of the methodology for local clients, and the rest of the moving pieces — keywords, reviews, links, citations — are covered in the series, starting with what local SEO actually works in 2026.

No promises in this post, and none to make: a unique city page is a necessary condition for the page to be indexable, citable, and worth having. Nothing about it is sufficient on its own.