An estate agent's site is two kinds of page with opposite lifespans. Listings arrive from a feed, live for a few weeks and go. Area pages — 'houses for sale in Didsbury', 'flats to rent in Leith' — should be permanent, and they are the pages that earn impressions year after year. Most agents' sites are built around the first kind and generate the second as an afterthought, which is backwards.
The feed is where the findings come from. Listings arrive with the portal's description, the same title as every other agent's copy of the same property, and leave by silently returning 200 with 'this property is no longer available'. Every one of those is a template decision, and the template is yours even if the data is not.
Measurement through a moving inventory is possible because the area pages do not move. Change the area template for some areas, hold others, and the comparison is between pages that are still there four weeks later — which listings, by design, are not.
The findings that recur on this kind of site
- Sold and let listings still returning 200Thousands of thin pages saying a property is gone, still in the sitemap, still crawled. Decide a policy — redirect to the area page after a grace period, or 410 — and apply it in the feed handler.
- Portal descriptions duplicated across every agentThe same three paragraphs on your site, the portal and four rivals. Duplicate content that Google resolves in the portal's favour, every time.
- Area pages that are a search result with a heading'Houses for sale in Didsbury' as a filtered listing and nothing more. Thin, and indistinguishable from a facet page, on the URLs that should carry the most.
- Search filters indexed as pagesBedrooms × price × type × area, each a URL with a canonical to itself, duplicating the area pages they should point to and spending the crawl on nothing.
- Branch pages with no structured dataMissing RealEstateAgent schema for the offices, on the pages that should tell the map pack and the assistants where you are.
What to change first, and what to compare it against
- Build the area pages first, and make them yoursLocal knowledge the portal does not have: the schools, the streets, the price history you have seen. That is the fix with the most durable return, and it ranks first by impressions once the listing noise is removed from the list.
- Fix the listing lifecycle in the template, onceExpired listings leave the index; live ones get an agent-written first paragraph above the portal copy. One change to the feed handler, applied to every listing that ever arrives.
- Use area pages as each other's controlsImprove half the areas, hold half, matched by traffic. The control-group method works here because the unit of comparison is the page that persists, not the listing that leaves.
- Read the window through the market, not against itInventory swings seasonally and with rates. Four weeks against a control over the same dates removes the market; four weeks against last month does not.
Questions from people running one
Should listings be indexed at all when the portals already rank them?
Live listings with an agent-written paragraph and photographs the portal does not have, yes. Listings that are the portal's copy verbatim, no — you are competing with a stronger copy of your own page. Many agents do best indexing the area pages and the branch pages and letting listings be noindexed but linked, so a searcher who arrives at the area page can browse them.
The feed changes daily. How can anything be measured?
By measuring the pages that do not change. Area pages, branch pages and guides persist through the inventory's churn, which makes them the experimental unit. A tool that watches those and ignores listing churn can produce a verdict; one that counts every expired listing as a new finding cannot.
Everything above is the method, usable with any crawler and a Search Console export. Rankelle runs it on a schedule: the crawl, the grouping into causes, the watch for the fix landing, the frozen control set and the verdict four weeks on — including the verdicts that say the work did nothing. One site is free, permanently, because the first verdict takes six weeks to arrive.
Try it on your own site, free
The whole audit of every website you add, no card and no expiry. Pay only when you want the proof.