The sitemap is mostly listings and tag pages, and the pages that should do the selling are well linked but hard to read
We looked at the 1,528-URL sitemap of example-property.co.uk across 12 sub-sitemaps, and at a 231-page crawl. Measured on 8 July 2026, prepared by Robin Laires.
Composition: 12 sub-sitemaps, 1,528 URLs
| Bucket | URLs | What it is |
|---|---|---|
| Property listings | 617 | Live for-sale and for-let stock. Indexable, but gone the day a sale completes |
| Blog posts | 550 | The content library, back to 2012 |
| Tag pages | 215 | Close to one tag archive for every two posts. Thin and overlapping, and they compete with the posts for the same queries |
| Core pages | 53 | Homepage, /sell/, /buy/, /fees/, /valuation/ and the guide steps. These should be doing the selling |
| Team profiles | 45 | One page per person. Genuine E-E-A-T raw material, currently thin |
| Property-type hubs | 22 | Apartment, bungalow, cottage and similar collection pages |
| Category archives | 9 | Blog category listing pages |
| Groups | 9 | Theme-generated grouping pages |
| Author pages | 4 | One per named contributor |
| Careers pages | 2 | Recruitment listing |
| Branch page | 1 | The single office location page |
| Property-area page | 1 | A single area or location landing page |
Much of the sitemap is short-lived or thin
617 of the 1,528 URLs are property listings, and 215 more are tag pages sitting over just 550 posts, roughly one tag archive for every two and a half articles. A listing disappears the moment the property sells, and a tag archive rarely says anything a post or category page does not already say better. Between them they make up the bulk of the sitemap.
Any ranking a listing earns resets to zero at settlement, so the visibility does not build up. The tag pages use up crawl budget and thin out the attention Google and the answer engines give the site, without adding anything new to index. The organic figures tell the same story from another angle: 356 of Harbour & Key's top 500 ranked keywords sit on /property/ listing pages that expire (see Google search).
The core 53 pages are well linked, and linking is not the problem
| Check | Result | Detail |
|---|---|---|
| No H1 heading | 42 / 53 | Includes /sell/, /buy/, /fees/, /valuation/ and every guide step. Nothing on the page tells Google or an answer engine what it is about |
| No meta description | 52 / 53 | The one-line summary shown in a search result, often quoted by AI answers, is missing almost everywhere |
| Empty title tag | 8 / 53 | The builder partnership pages (/barratt/, /taylor-wimpey/ and similar) |
The core 53 pages sit one link layer deep. The crawl found no orphans within this sampled set, and Google and the answer engines can reach every one of them without difficulty.
The gap is what the pages say once they are found. A page an engine can reach but cannot read (no H1, no meta description, and in eight cases no title at all) is close to a page it cannot reach. It is good wiring into empty rooms.
The property-type hubs look like wasted potential
22 property-type pages exist (apartment, bungalow, cottage and similar), and some carry collection-page schema, which tells Google they are a curated set. But the crawl found three pages with an empty title tag inside this bucket. The three pages the crawl could not find any internal link to, anywhere in the sampled set, were also here: /property_type/bungalow/, /property_type/detached-bungalow/ and /property_type/detached-house/.
A property-type hub is the kind of page that could outlast any single listing and become a lasting entry point for searches like “bungalows for sale in Edinburgh”. Right now these hubs have no title and, for a handful, no link to them. The lasting version of what the listing pages do today already exists as a URL. It is switched off.
The blog answers the same question on new pages every few years
| Topic | Pages | Where |
|---|---|---|
| Offers over | 4 pages | Posts from 2011, 2022, 2023 and 2026 |
| Home reports | 3 pages | A guide step plus 2024 and 2025 posts |
| Notes of interest | 3 pages | A guide step, a legal page and a 2023 post |
| Additional Dwelling Supplement | 2 pages | Posts from 2017 and 2018 |
Across 14 years of posting, the same seller questions have been answered again on new URLs, and one canonical page has not been kept up to date. The ranking data shows the examples above. The /buy/ guide and /legal/ section also duplicate each other on notes of interest, submitting an offer and finding a solicitor.
When several Harbour & Key pages chase one question, Google splits the relevance signal between them and may rank none of them well. The home report cluster is the measurable cost: real search volume, three competing pages, and rankings at #14 to #17 and not on page 1. One strong page per question, with the dated versions redirected into it, could move that cluster. Branded search is the one place this works in the site's favour, where two Harbour & Key pages hold #1 and #2 for “harbourkey edinburgh”.
Six fixes, in order
| # | Fix | Why |
|---|---|---|
| 1 | Noindex the 215 tag pages | They add nothing a post or category page does not already provide, and removing them from the index stops them competing with the content they sit over. |
| 2 | Keep the 617 property listings, and stop relying on them for the SEO load | They are the right pages for a buyer browsing live stock. They are the wrong foundation for rankings that need to survive a sale. |
| 3 | Build suburb and area hub pages that persist (Morningside, Leith, Marchmont and similar) | They are the lasting version of what a listing page does today, and they can absorb the search demand a sold listing leaves behind. |
| 4 | Give the 22 property-type hubs a working title tag and an internal link | Starting with the three the crawl could not find a route to at all. |
| 5 | Restore H1s, titles and meta descriptions on the 53 core pages | These pages are already well linked and are waiting to be readable. |
| 6 | Consolidate the duplicate-topic pages into one page per question | Merge the four offers-over posts into one, the three home report pages into the guide step, and the duplicated buy-guide and legal pages into a single URL each. 301 the dated versions into the survivor. |
The full sequence is on the action plan.
Counts come from the 1,528-URL sitemap across 12 sub-sitemaps and a 231-page crawl, run on 8 July 2026. The crawl reads the code each page first sends. The orphan check covers the sampled core 53 pages only.