The boring, specific ones — “water heater repair [neighborhood],” “emergency AC repair [city],” “furnace repair near [landmark]” — bring paying customers more reliably than broad head terms like “best [service] in [city].” That is what local-SEO practitioners report from their client accounts, and we label it as an observational finding, not a controlled study: no one has randomized this, but the pattern repeats independently across local work, and it matches what customers describe doing when they hire.

The practical consequence is that a local site’s keyword plan should start from the bottom of the funnel and work up, not the other way around. This post is part of the local-SEO series (the overview is what local SEO actually works in 2026) and covers four things: why the head term is a trap, what the bottom-funnel queries actually look like, the competitor-gap method for finding them in about an hour, and why the same pages are the ones AI answers can cite.

Why “best [service] in [city]” is a trap

The head term has two problems, and both work against the business.

Competition. Every other business in the city targets the same query, plus the review platforms, the listicle publishers, and the aggregators who rank for “best” terms by design. A small local business asking a website to win that SERP is asking it to beat its own competitors and the businesses whose entire model is to own comparison content.

Intent. The person typing “best plumber in [city]” is usually comparing, not ready to commit — and often not in the moment of need. The click, when it happens, frequently goes to a competitor or a review platform, and the customer arrives with a shortlist that does not include you. The person typing “water heater repair east [city]” is one step from making a call.

Neither effect is a law; it is the consistent pattern practitioners report. The operational point is that optimizing the head term first spends your effort where the competition is densest and the intent is thinnest.

What bottom-funnel local queries look like

The pattern: specificity is the signal. The more specific the query — the service named, the place named, the urgency named — the closer the searcher is to a phone call. Below, three broad and three specific queries from a plumbing/AC account, with intent and competition notes. The competition ratings are illustrative judgments for this example, not measurements.

QueryTypeWhat the searcher is doing (intent)Competition (illustrative)
“best plumber in Springfield”BroadComparing options; not yet in the moment of needHigh — every local competitor, review platforms, listicles
“plumbing services Springfield”BroadEarly research, maybe scoping a projectHigh — category pages, directories, competitors
“what does a water heater replacement cost Springfield”Broad-ishPricing research; may be weeks from a callMedium — fewer local pages, more national content
“water heater repair east Springfield”SpecificSomething is leaking right now; calling todayLow — few pages target the neighborhood at all
“emergency AC repair Springfield”SpecificNo heat/cold, in crisis, calling nowLow–medium — urgency pages exist but are often generic
“furnace repair near [arena/riverfront]”SpecificLocal, directional, often mobile, in the momentLow — landmark-anchored pages are rare

Two things to notice. First, the specific queries are ones a generic marketing agency never asks the client for, because they don’t look like “keywords” — they look like a customer talking. Second, the competition column is where the strategy lives: the bottom-funnel set is winnable for a small business with a well-built city page, and the broad set usually isn’t, at least not at the top.

This is also why a rank tracker alone misses the point: it measures whether you hold the queries you were given, not whether those queries were ever the ones that convert. The query list has to be built from intent, not from whatever a tool says has volume.

The competitor-gap keyword method, step by step

You do not need expensive tooling to build the first version of this list. The method below takes roughly an hour and produces a concrete set of pages to build or fix. It is a practitioner standard rather than a research result; we present it as a working method, not a formula.

  1. Pick the two or three competitors you most want to be. Not the biggest — the one whose phone you wish was ringing. Usually the business that owns the top of the local results panel for your money queries.
  2. List what they already claim. Their site is a target list in itself: every city, service, and service+city page they have published is a query they are targeting. Note which of those you have no page for.
  3. Check search results directly. For your top services, run the variants — “service in [city],” “service [neighborhood],” “emergency service [city],” “service near [landmark]” — in a clean browser. Note which of your three competitors appears in the top five, and which does not. This is a manual SERP check, and it is enough for the first pass.
  4. Use Search Console access where you have it. If the client (or an agency with the client’s consent) has Google Search Console access, the queries page is the cleanest version of this: which “service + place” queries already return impressions and low positions are gaps on your own property. Without access, step 3 substitutes.
  5. Build the gap list: competitor present, you absent. Rank the gaps by intent — bottom-funnel first: urgency (“emergency,” “24 hour,” “same day”), then geography (neighborhood, landmark), then cost and decision queries (“cost,” “warranty,” “recall”).
  6. Assign a page to each target. Each gap gets a matching page — the city+service structure from the one-page-per-city post, with real local detail. A gap with no page behind it is a wish.
  7. Re-run the SERP check monthly. As pages land and age, re-run the same query list and note movement. The list of gaps shrinks or changes shape; the check is what tells you when.

Steps 2–4 are the part most local keyword plans skip, because they start from the client’s brand or from a tool’s keyword export instead of from what competitors already prove works in this market. The market is already telling you which queries convert — the competitors’ pages are the receipts.

What the exact-match city page looks like

For a gap like “water heater repair east Springfield,” the page is an exact match: the query is the title, the URL encodes the place and service, and the body is real — named neighborhoods, specific jobs, local reviews, a direct phone number in the first screen. The mechanics and the anti-pattern (templated city pages) are covered in depth in do you need a separate page for each city you serve.

One caution on the evidence, stated plainly: a local practitioner has publicly described building exact-match city pages for his own business and reporting #1 rankings in the city’s search results. That is a single, self-reported result on one site in one market — an anecdote. It is directionally consistent with everything in this post, but it is not proof, and we would not build a plan on it. What the anecdote does add is the reminder that these pages work for the person who builds and maintains them, not for an agency that ships them once.

Why the same pages serve AI answers

The bottom-funnel query pattern has a second consumer, and it is not the customer’s phone. When a customer asks an AI surface a local question — “who fixes water heaters near [neighborhood],” “emergency plumber in [city],” “AC repair open now” — the system grounds the answer in pages it can retrieve, select, and cite. The selection logic is unpublished, but the requirement is visible in what gets cited: specific, concrete, checkable content about a named place. A “best [service] in [city]” page with generic copy has almost nothing quotable. A city page with named neighborhoods, dated jobs, and local reviews is exactly the shape an answer can draw on.

The per-surface mechanics — how each major surface retrieves and cites, and how to check your own brand on each — are worked through in which AI surfaces cite B2B brands, and the general pipeline and levers are in our AEO and GEO one-page primer. The local version of the extractability lever is unfashionably simple: the page must say the place and the specifics in plain text, so both Google and an answer engine can lift it.

There is a compounding effect worth naming honestly, as a hypothesis rather than a law: pages that win the bottom-funnel search click are the same pages that are citable in AI answers, so a well-built city page earns in both channels at once — and the review content that feeds one feeds the other. We have not measured that compounding; the structure implies it, and practitioners anecdotally report the overlap.

If this is your situation

If your current local keyword plan is a spreadsheet of “service in [city]” terms with volume numbers, the next step is the hour-long audit above, not more keyword volume. Build the gap list from the two or three competitors you most want to be, rank it by intent, and assign a page to each gap — the city+service structure with real local detail. The rest of the local stack that has to be in place for those pages to hold — the review system, and the links and citations that corroborate the business as a place — are covered in the series, including the local links that actually work, the citations and NAP cleanup and the review-velocity system. We run the full audit, keyword gap through page build, for local clients under the methodology.

No promises in this post: “these keywords convert better” is a practitioner pattern, labeled as such. What is not labeled is the structural part — the query list built from intent, and the pages that answer it — because that part you can build and inspect yourself.