The problem with a search box
A rental platform puts everything behind a search box. Google cannot type into a search box, and neither can somebody who has just asked their phone what a flat in one particular neighbourhood costs. Every one of those searches was landing on a homepage that answered it with a form.
So we stopped waiting for the crawler to guess, and built the answers as pages.
What we built
One page per neighbourhood, generated from the catalogue itself rather than written by hand. Each one opens with the sentence somebody was searching for — how many flats are available there, what the median rent is, what the usual range and size are — and then the listings underneath it.
They are not static. The catalogue changes every day, so the pages are rebuilt every night from the same data the site serves; a neighbourhood that has three flats today says three, and a neighbourhood that has none says so rather than showing an empty grid.
- ✓A page per zone, generated — adding a neighbourhood is a row, not a rewrite
- ✓The count, the median rent and the median size on the page, not in a database
- ✓BreadcrumbList, FAQPage and RealEstateAgent structured data on every one
- ✓Three real questions answered per zone, in the markup search engines read
- ✓An
llms.txtat the root, because the thing reading the site is increasingly not a browser - ✓Everything in one sitemap, submitted and monitored in Search Console
What is on the site right now
These are counted from the live sitemap and the live pages as this was written. Open them and count again.
A worked example: the Astir page currently opens with sixty-three flats, a median rent of €300 and a median size of 68 m² — and the smaller neighbourhoods say five, fourteen, sixteen, because that is what is there.
What we are not claiming
Rankings belong to Google, not to us, and they move. Search Console is connected and the sitemap is submitted, so the position and impression data exists — but it is the client’s data, it changes weekly, and a number screenshotted once and printed on an agency page is worth nothing to you.
What is checkable is on this page: the pages exist, they carry the right structured data, the numbers on them are the real ones, and they were rebuilt last night. That is the part we did.
The same method, on a smaller site
This does not need six hundred listings. A dental practice with eight treatments, a florist delivering to twelve postcodes, a solicitor with four practice areas — each of those is a set of real questions that currently has no page to answer it. The work is the same: find what people type, generate a page that answers it from data you already hold, and mark it up so a machine can read it.