The map is two columns, the query and the page that owns it. Barnoldswick Baths fits bathrooms across the Yorkshire and Lancashire border and had 180 queries from a morning of harvesting and 26 pages on the site. Mapping the first onto the second took two hours and found things nobody would have found by writing.
180 queries mapped onto 26 pages, and 9 pages had no query at all. Nine pages on a small site existed because somebody had once thought they should, answering nothing anybody types. A page with no query is a page nobody asked for, and the map is the only cheap way to notice.
What a row looks like
One page owns one primary query and as many secondary ones as share its intent. The wet room page owns wet room installation yorkshire as its primary and carries eleven secondary queries in the same row: wet room cost, wet room vs shower enclosure, is a wet room a good idea in a small bathroom, and eight more. Secondary queries go in the same row, not in a new one, and that single rule prevents most of what would otherwise become two pages fighting.
The map
Four rows of a map, with the queries each page owns
Every query sits under exactly one page, and the page with nothing under it is the finding rather than the mistake.
26 pages, 180 queries
- The wet room pagePrimary: wet room installation yorkshire
- wet room cost
- wet room vs shower enclosure
- is a wet room a good idea in a small bathroom
- The drainage pageA different person with a wet room already
- wet room drainage falls
- wet room floor leaking
- The bathroom fitting costs pagePrimary: bathroom fitting cost yorkshire
- how long does a bathroom take to fit
- bathroom fitting labour only price
- The company history pageOne of the nine pages with no query under it at all
The fourth row is kept on purpose, because it exists for customers rather than for search. Three other pages like it were removed, and the map is where that decision is recorded.
The test for whether a query belongs in a row or needs its own is not the words. It is whether one page can satisfy both askers without either of them scrolling past a section written for somebody else. Wet room cost belongs with wet room installation; wet room drainage falls sits with a different page, because the person asking has a wet room already and a problem with it.
The fourteen commissions
A query with no page is a commission, and there were 14 of them. That is the output most people expect from this exercise and it is the smaller half of the value. Fourteen rows with an empty right hand column is a year of writing, ordered by nothing yet, and each one already carries the queries it has to answer.
Audit outcomes
What empty cells on both sides of the sheet became
Mapping left fourteen queries without a page and nine existing pages without a query, splitting the work into four actions.
- Commission new pages2
- Write14 queries with an empty page column
- Forms the writing backlog with target queries attached
- Merge2
- Fold inTwo unmapped pages
- Combined into existing pages that owned queries
- Keep intentionally2
- LogFour customer-facing pages
- Kept for buyers, recorded so nobody re-audits next year
- Remove2
- RedirectThree unneeded pages
- Removed and pointed to relevant pages
Recording why a non-search page stays prevents teams from rediscovering and questioning it a year later.
The nine pages with nothing in the left hand column got three different verdicts. Two were merged into pages that did have queries. Four were kept anyway because they exist for customers rather than for search, which is a legitimate answer and worth writing in the sheet so nobody rediscovers them in a year. Three were removed and redirected.
Where this stalls: mapping after the writing
Write the map before the pages, because it is cheaper to move a row. A row takes a second to reassign. A page that has been written, linked and indexed for a query it should never have owned takes a merge, a redirect and three months.
The stall
Moving a row before writing versus fixing a published page
Teams skip mapping because a spreadsheet feels invisible, until two live pages collide and force a three-month cleanup.
Barnoldswick only built its map after two live pages had already collided on the same query.
The stall is that mapping produces nothing visible. There is no page at the end of the afternoon, nothing to show anybody, and the work is a spreadsheet, which is why it is the step almost every small site skips and every large site has been forced into by pain. Barnoldswick only did it because two of its pages had already collided, which is two of your pages competing for one query and is the ordinary route to caring about this at all.
Using it afterwards
Check the map before any new page is written. That is the whole discipline and it takes thirty seconds: search the sheet for the query, and if a row exists, the work is an edit to that page rather than a new one. Barnoldswick has caught four proposed pages this way since March 2026, three of which became sections.
Governance
The thirty-second check before any new page is drafted
Searching the sheet first decides whether a topic becomes a section on an existing URL or a brand-new row.
- Yes, a row existsEdit the existing pageExpand the owning URL instead of launching a second competing page.
- Caught four proposed pages since March 2026
- Three became new sections on existing pages
- No row exists yetAdd a row, then commissionAssign the query, target URL, and last-reviewed date in Google Sheets.
- Not a sitemap of live URLs or a dated schedule
- Third date column drives an annual content review
Kept free in Google Sheets; writing priority is decided separately against demand and effort.
The map is not a sitemap and it is not a plan. A sitemap lists what exists and says nothing about what each page is for. A plan has dates in it and this has none, deliberately, because the order of writing is a separate decision made against demand and effort rather than against the map.
Google Sheets holds it, free, one row per page, and it replaced nothing because there was nothing before it. The two columns that matter are the query and the address; the third useful column is the date the page last answered the query well, which turns the map into a review list once a year.
Where the 180 came from is sources, and building the first list, and deciding which queries share a row is what the searcher wants, which the words hide applied one row at a time. The map also describes a shape the site does not yet have, and building that shape is a hub, its supporting pages, and the links between. All of it is the deliverable of finding what people type, and what they mean by it, and it is the one nobody can show anybody.