Local keyword research and service-area pages that do not cannibalise

Local keyword research looks like normal keyword research with a place name bolted on. It is not, because local queries resolve into two different result layouts - the map pack and the organic listings - and those layouts reward different pages.

Get this wrong and you build a beautiful city page for a query that only ever returns a map pack, where a web page cannot rank at all.

The four local query patterns

• Service plus city - "emergency electrician manchester". High commercial intent, usually a mixed map pack and organic result. This is where service-area pages earn their keep.
• Near me - "dentist near me". Almost purely a Business Profile query. Your page copy barely participates; distance and prominence decide it.
• Implicit local - "emergency electrician" typed from within the city. Google localises it silently. Same treatment as service plus city.
• Local informational - "how much does rewiring cost uk", "do i need building regs for a loft conversion". No map pack, pure organic, and consistently the cheapest local traffic to win. Most local businesses ignore it entirely.

Check the pattern before you write. Search the term, look at what the page returns, and let the layout tell you which asset to build.

Building the keyword map

Start from your actual revenue lines, not from a keyword tool. List every service you sell and every location you genuinely serve, then cross them. That grid is your candidate set.

Now prune it, hard. A location belongs on the list only if you can say something true and specific about serving it. If you cannot name a project, a travel time, a local constraint, or a relevant regulation, you do not have a page - you have a template.

Volume data for local terms is unreliable at the tail, so rank the grid by commercial value rather than by reported searches. A term with twenty searches a month and a four-figure job behind it beats a term with a thousand searches and no purchase intent.

Why most service-area pages fail

The standard failure is the spun page: one template, the town name substituted, published forty times. It fails for two independent reasons.

First, it is thin duplicate content, and Google will simply pick one of the forty and ignore the rest - or none of them. Second, it converts badly, because a visitor can tell within a sentence that nobody wrote this for their town.

The tell is almost always in the title tags. Forty pages that differ only by a place name usually end up with near-identical titles and descriptions, and duplicate titles across a site are one of the clearest signals that pages are competing with each other rather than covering distinct ground.

What a service-area page needs to contain

A page earns its place when it holds information that exists nowhere else on your site. In practice that means:

• Work you have actually done in that area, named and described.
• The specific constraints of the location - parking, access, building stock, common faults, local regulations.
• Genuine local proof: reviews from customers there, photos from those jobs, the names of neighbourhoods you cover.
• Pricing or timing that reflects that area, if it genuinely differs.
• A clear, single call to action matched to how people in that market buy - phone for emergency trades, form for considered purchases.

If you cannot fill that out for a location, do not publish the page. One strong city page plus a well-structured services page will outperform fifteen thin ones, and it will not put the rest of your site at risk.

Structuring the site so pages do not collide

Two structures work. Service-led - /services/rewiring/ with location detail inside - suits businesses with a small service area and many services. Location-led - /manchester/rewiring/ - suits businesses with many locations and few services.

What does not work is running both at once, because then /services/rewiring/ and /manchester/rewiring/ compete for the same query and split the signals. Pick one axis as primary and make the other a filter or a section, not a parallel URL tree.

The informational layer everyone skips

Local informational queries have no map pack, which means the field is open to whoever writes the best answer. Cost guides, regulation explainers, seasonal problems, and "should I repair or replace" comparisons all pull qualified local traffic and build topical authority that lifts the commercial pages alongside them.

They also feed the assistant layer. When someone asks an AI assistant what a job costs in their area, the sources it draws on are exactly these pages - and almost nobody in local services is writing them properly.

Pages get you into the organic results. Getting into the top three of the map pack is a different game, decided largely by prominence - which is the next lesson.