Where "five pages" comes from
It comes from pricing, not from search.
Five pages is a convenient unit of work to quote. It is roughly a week, it fits a fixed price, and it describes a business rather than answering any particular question. Nobody arrived at it by looking at what customers type.
And it was a reasonable default once, when a local trade competed against the handful of other local firms who also had a website. It stopped being reasonable the moment search started matching a specific query to a specific page — which is to say, a long time ago.
The failure it produces is subtle, which is why it persists. The site is not broken. It looks fine, it loads, the form works. It simply never appears for the searches that would have brought work, and there is nothing in any report to tell you which ones those were.
The arithmetic that gives you your number
Work it out rather than picking it. Four things to count.
Services that are searched separately
Not your internal service list — the things customers type as distinct phrases. A paving company does driveways, patios, block paving, resin, tarmac, garden walls and drainage. A homeowner searching for a resin driveway is not searching for a patio, and one page cannot be the best answer to both.
Places you genuinely work
The towns where you have done jobs, would travel again, and can describe. Not every place name within thirty miles — the towns with real work behind them, so the page can name streets, describe the housing stock and show a job you actually did there.
Questions asked before anybody rings
How long does a resin driveway last. What does block paving cost per square metre. Do I need planning permission for a dropped kerb. These are searched months before the enquiry, and the firm that answered them is the firm that gets the call.
The pages that close it
Guarantee terms, how payment works, how you handle access and mess, insurance, who actually turns up. Boring, and read carefully by anybody about to spend four figures.
You do not multiply services by places exhaustively — that way lies four hundred thin pages and a doorway problem. You build the combinations that have real search behind them and real work to describe. The honest test for each page is whether you could write six hundred specific words about it. If you cannot, it should be a section on a broader page instead.
What twenty-seven actually looked like
Pro Pave is a paving company competing for high-value domestic work across Greater Manchester, and a client on this site. The build came to twenty-seven pages of services and locations.
It was not a target and nobody was padding to reach it. It is what happened when the service list was broken into the things people search for separately, and crossed with the towns where there was genuine work to point at.
The reasoning was the same on every page: each search needed somewhere of its own to land rather than everything competing with the homepage. A homepage that mentions driveways and patios and block paving across five boroughs is one page with a diluted claim to all of them.
Two things about that build are worth separating out, because people assume more pages means a slower, heavier site. They do not have to be related at all.
The site is a bespoke WordPress theme rather than a page-builder assembly. The homepage answers in 51 milliseconds, shows content in 0.56 seconds, loads completely in 0.68 seconds, renders from eleven requests, and has a cumulative layout shift of zero — measured on the live site on 4 September 2026 using the browser's own Navigation Timing, Resource Timing and Layout Instability APIs.
Eleven requests on a twenty-seven page site is the point. Page count is a content decision; page weight is a build decision. Conflating them is how trades end up with five slow pages instead of twenty-seven fast ones.
The version of this that goes wrong
There is a bad version of everything above and it is worth naming, because plenty of agencies sell it.
The bad version is a template page duplicated across forty town names with the place swapped in. "Driveways in Bury. We install driveways in Bury. Our Bury driveway installers..." — the same six paragraphs, forty times, with a find-and-replace.
Google has been acting against that pattern for years, it is explicitly what the doorway-page guidance describes, and the risk is not confined to the thin pages. A cluster of them drags on how the whole site is assessed, which means forty bad pages can cost you the four good ones.
The difference between the good version and the bad one is whether the page has anything of its own on it. A real town page names jobs done there, shows photographs taken there, mentions the kind of properties and the kind of problems common in that area, and reads as though somebody who has worked there wrote it. If you cannot do that for a town, you have not earned a page for that town yet — and the honest move is to wait until you have.
This site applies the same test to itself. Of forty-two towns measured in our own research, eighteen have a page. The rest sit in a table on the borough page that covers them, because there was not enough to say.
The order to build them in
You do not need all of them at launch, and trying to write twenty-seven pages before going live is how a site takes nine months.
A reasonable order:
- The highest-margin service you want more of, as a proper page. One page, done well, for the work you actually want.
- The town where you already do most of that work. You have the photographs and the stories, so it writes itself.
- The phone number tracking, before anything else is added. Without it you cannot tell which of these pages is working, and every decision after this point is a guess.
- The two or three questions customers ask you on every single call. You already know what they are — you answer them daily.
- Then the next service, the next town, one at a time, led by what the tracking says is producing calls.
That order means the site is earning from month one and growing in the direction the data points, rather than being guessed at wholesale and then abandoned.
It also means the page count is never really a question you answer in advance. You answer it by building until the additions stop producing calls.
Working out your own number
- List the services customers search for separately, in their words rather than yours.
- List the towns you have genuinely worked in and would travel to again.
- For each service-and-town pair, ask whether you could write six hundred specific words and show a real job. If not, do not build it yet.
- Add the questions you answer on the phone every week.
- Add the guarantee, payment, access and insurance pages.
- Count. That is your number, and it is almost certainly not five.
- Check the build separately: how many requests does your homepage make, and how long does it take on mobile data? Page count and page weight are different problems.
If the number comes out at eight, build eight. The figure is not a virtue — the point is that it came from counting searches rather than from a package.
