Nothing blocks Google or the AI crawlers, and phone speed looks like the main thing to work on
We tested the Harbour & Key homepage on a phone and on desktop, checked AI crawler access and ran a 231-page crawl of example-property.co.uk. Measured on 8 July 2026, prepared by Robin Laires.
- What we found
- On a phone, the main content of the homepage took 16.7 seconds to show. On desktop it took 2.0 seconds. Google and the AI crawlers can read the site.
- Why it matters
- Most property searches start on a phone, and Google's ranking systems look at this experience.
- What could help
- Trimming the scripts and styles a phone does not need for the first screen, and checking two legacy addresses that lead nowhere.
Access is open
GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot all get a 200 response, the same as a plain browser and the regular Googlebot user agent. robots.txt is open and the sitemap is declared. The desktop score of 82 / 100 and the technical SEO score of 92 / 100 are also sound. From what we can see, the engines can read everything on the site. The gap is the amount on the money pages for them to lift (see content and trust and the scorecard).
Scores and checks on the homepage
| Check | Result | Status |
|---|---|---|
| Technical SEO score, homepage (mobile) | 92 / 100 | good |
| Technical SEO score, homepage (desktop) | 92 / 100 | good |
| AI crawler access (GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot) | 200 on all four | good |
| robots.txt | Open, sitemap declared | good |
| llms.txt | Not present | good |
| Mobile performance score | 44 / 100 | critical |
| Desktop performance score | 82 / 100 | good |
Main content takes 16.7 seconds to appear on a phone
It is the same server, the same page and the same content on both. That points to the build and not to hosting. Most property searches start on a phone, and Google's ranking systems measure exactly this experience.
What could help: cutting the 746 KiB of unused JavaScript and 144 KiB of unused CSS the mobile run carries. A first milestone of main content under 4 seconds could be set (Phase 1 of the action plan), then pushed toward 2.5 s.
Homepage, mobile and desktop, 8 July 2026
| Metric | Mobile | Desktop | Target | Status |
|---|---|---|---|---|
| Performance score | 44 / 100 | 82 / 100 | 90+ | critical |
| Main content visible (largest contentful paint) | 16.7 s | 2.0 s | under 2.5 s | critical |
| First paint | 8.3 s | 1.4 s | under 1.8 s | high |
| Main thread blocked (total blocking time) | 440 ms | 20 ms | under 200 ms | high |
| Layout shift | 0.029 | 0.012 | under 0.1 | good |
746 KiB of unused JavaScript and a 4,033 KiB page
| Signal | Value | Status |
|---|---|---|
| Unused JavaScript, mobile run | 746 KiB | high |
| Unused CSS, mobile run | 144 KiB | medium |
| Page weight, mobile | 4,033 KiB | high |
The mobile run downloads and parses 746 KiB of JavaScript that never gets used, plus 144 KiB of unused CSS, on a page that weighs 4,033 KiB. From what we can see, that is the direct cause of the 16.7 second wait and the 440 ms of blocked main thread.
What could help: trimming or deferring the scripts and styles the first screen does not need, then re-testing against the 4 second milestone.
Every crawler we tried gets a 200
| Crawler | Response | Status |
|---|---|---|
| GPTBot | 200 | good |
| OAI-SearchBot | 200 | good |
| ClaudeBot | 200 | good |
| PerplexityBot | 200 | good |
| Google-Extended (regular Googlebot user agent) | 200 | good |
| Plain browser | 200 | good |
There is no llms.txt file, and the evidence here suggests that does not matter, because none of the engines need it to read the site. Nothing to fix here, and adding llms.txt would not change anything we measured.
Two legacy paths and one subdomain lead nowhere
| URL | Response | Note |
|---|---|---|
| /dvd/disc_show.asp | 404 | 7 old inbound links point here |
| /dvd/freetalk_show.asp | 404 | 4 old inbound links point here |
| enquiries.example-property.co.uk/valuation | 403 to every client tested | 7 external links point here |
The old /dvd/*.asp addresses are left over from a previous site build. They now return a 404, and 11 old inbound links still point at the two of them. The valuation subdomain returned a 403 to every client tested in this audit, while 7 external links point at it. The /valuation/ page on the main domain loads fine (200), so the subdomain could be an over-tight firewall rule or a dead end from an old funnel. It would need a look from inside to say which. More on this on the enquiries and tracking page.
What could help: a 301 from the /dvd/*.asp addresses to their current equivalents, and a check of the firewall rule on the enquiries subdomain. This looks like an afternoon of work (Phase 0 of the action plan).
The automated checks pass, and that is only part of the picture
The technical SEO score is 92 / 100 on both mobile and desktop, HTTPS is fine, and the site runs on WordPress with Yoast. That score covers basics such as crawlability and viewport. It does not cover content quality. The same audit found that 42 of the 53 core pages carry no H1 heading and 52 of 53 have no meta description (see the scorecard and site structure). A clean 92 here should not be read as the money pages being in good shape. Restoring the H1 headings and meta descriptions on the 53 core pages is tracked in Phase 0 of the action plan.
We ran Google's page performance test on the homepage, phone and desktop, on 8 July 2026, and crawled 231 pages. AI crawler access was checked on the same date. Speed figures come from one test and will vary between runs.