08 · Content and trust

The content library and the reviews look like real strengths, and the gaps are mostly formatting

We looked at the blog library, on-page formatting and the trust signals AI systems already read on example-property.co.uk. Measured on 8 July 2026, prepared by Robin Laires.

The short version
What we found
About 550 posts back to 2012, guides that already rank, and 4.6 from 662 Google reviews. 60 of 159 sampled posts have no meta description, and nothing is shaped as a direct answer.
Why it matters
Google and AI systems write their own summary when a page gives none, and they tend to lift text that already reads like an answer.
What could help
Adding meta descriptions, short direct answers and question-style headings to the pages that already rank.
~550
Posts in the content library, back to 2012
60 of 159
Sampled posts with no meta description
14
H1 headings on the 2014 LBTT guide, still ranking
4.6 / 662
Google rating and review count, the largest we measured
45
Team pages, around 600 words each
0
Thin pages in the 231-page crawl
Finding 1

The library looks sound, and the formatting is what holds it back

good
Ranking on real search volumePositionSearches a month
best areas to live in edinburgh#17590
LIFT scheme guide#16720
offer in region of meaning#11Not stated
Home report questions cluster#14 to #17Not stated

Roughly 550 posts back to 2012. Guides run 1,000 to 1,900 words (the buying guide steps run 1,251 to 1,886 words and the selling guide steps 1,081 to 1,640), and the 231-page crawl found no thin pages. The seller-education journey looks largely written.

What could help: launching any new guide at the same depth as the existing library, and linking new content to the pages that already rank.

Finding 2

60 of 159 sampled posts have no meta description

high

Just over a third of the sampled blog library has no meta description. When the field is blank, Google and answer engines write their own summary of the page. That is the line a seller reads before clicking, and the sentence an AI answer may lift to describe the page, and Harbour & Key does not control it.

What could help: writing a meta description for all 60, starting with the posts that already rank (the LBTT and home report clusters), since those already earn impressions.

Finding 3

The 2014 LBTT guide carries 14 H1 headings, and it still ranks

high

The strongest case in the crawl is a 2014 guide to the land and buildings transaction tax, with 14 separate H1 headings on one page. It is a live page. Related additional dwelling supplement (ADS) terms still rank 27 to 31 for Harbour & Key, at 590 to 720 searches a month. That suggests this content still has commercial value in its current state.

What could help: rewriting the page and keeping the URL. One H1, current LBTT and ADS rates, and a direct answer near the top. Deleting a page with live rankings could throw away traffic that took over a decade to earn.

Finding 4

Nothing on the site is shaped as a direct answer

high

Across the library, no guide opens a section with a short direct answer, and no H2 on a money page is phrased as the question a seller (or an AI system) would ask, such as “how much does it cost to sell a house in Scotland”. Answer engines lift text that already reads like an answer. When a page makes them assemble one from paragraphs, they tend to choose a source that already did the work.

What could help: a 40 to 60 word direct answer straight after the heading on the top guides, and question-style H2s on /fees/ and the guide steps. This is a formatting change and the underlying research stays as it is.

Finding 5

The trust layer AI systems already read

good

4.6 from 662 Google reviews, the largest review count of any Edinburgh agent we measured. The allagents.co.uk and Trustpilot profiles exist, and ChatGPT already builds its answers to branded questions about Harbour & Key from these sources (see AI visibility). The raw material for trust is in place. It is not yet wired into the money pages.

What could help: keeping review velocity up on Google, allagents and Trustpilot. These profiles already do real work inside AI answers, so losing momentum here would be a step backwards.

Finding 6

45 team pages look like untapped E-E-A-T material

medium

Harbour & Key has 45 team pages, real named people at a firm that is both agent and solicitor. They run thin at around 600 words, with no Person schema beyond the theme default. The expertise behind the £600 fee is barely legible to Google or an AI system.

What could help: building these pages out with qualifications, specialisms and a line on the solicitor-plus-agent model, and marking them up properly. Harbour & Key has agency and legal staff under one roof, and other agents may find that harder to say.

What an answer capsule looks like

The LBTT rewrite, before and after

The 2014 guide already earns rankings on ADS terms. This is the shape a rewrite could take: one H1, a question-style subheading, and a short direct answer before the detail, so Google and an AI system can lift the opening paragraph as it stands.

Before (current page)
H1: Goodbye Stamp Duty, Hello Land and Buildings Transaction Tax
H1: What changed in 2015
H1: How LBTT is calculated
... (14 H1s total, no single answer, no current rates)
After (one H1 plus a direct answer)
H1: What is LBTT and how much will you pay in Scotland?

Direct answer (40-60 words): Land and Buildings Transaction Tax (LBTT)
is Scotland's version of stamp duty, charged in bands on the purchase
price of a property. An Additional Dwelling Supplement (ADS) applies
on top for second homes and buy-to-lets. [Current bands and rates.]

H2: How is LBTT calculated on a single property?
H2: What is the Additional Dwelling Supplement (ADS)?
H2: Do first-time buyers get an LBTT discount in Scotland?
Method and limits

Figures come from the 231-page crawl, a sample of 159 blog posts and the ranking data, measured on 8 July 2026. Word counts and heading counts are read from the page code. Related pages: the site structure and the scorecard.

Real client work. Names and locations changed.