02 · Scorecard

Scorecard

We scored eleven areas from 0 to 4, using live checks on 8 July 2026.

The short version
What we found
Harbour & Key scores 22 of 44. Reviews and branded AI answers score full marks, and six areas score 1 of 4.
Why it matters
The low scores sit where a seller may look when choosing an agent: Maps, Google search and AI tools.
What could help
Adding page headings and titles and structured data, and a lighter mobile homepage, could help rows 6, 7 and 8. The Business Profile and review work could help row 1.
Overall
22 / 44
50%

Each of the eleven areas is scored 0 to 4 against what a strong local solicitor estate agent might look like, and the scores are added up with no weighting. Every score comes from a live measurement on 8 July 2026.

2
Areas at 4 of 4: branded AI answers, reviews
2
Areas at 3 of 4: content library, backlinks
1
Area at 2 of 4: search breadth
6
Areas at 1 of 4
0
Areas at 0 of 4
22 of 44
Points in total
Eleven areas, scored 0 to 4

Scores by area

Maps rankings
1 / 4
Branded AI answers
4 / 4
Seller AI answers
1 / 4
Searches that matter
1 / 4
Search breadth
2 / 4
Mobile speed
1 / 4
Headings and titles
1 / 4
Structured data
1 / 4
Content library
3 / 4
Backlinks
3 / 4
Reviews
4 / 4
Scores and what we saw

Eleven areas

#AreaScoreWhat we sawWhat good looks like
1Maps rankings1 / 4Harbour & Key is not in the top 20 for any of 5 live Maps searches: “estate agents” Edinburgh, Glasgow and Musselburgh, “solicitor estate agents” Edinburgh and “fixed fee estate agents” Edinburgh. Neilsons (4.8/194), Umega, Warners, McEwan Fraser and VMH (4.9/321) hold the packs, despite the 662 reviews at 4.6.In the 3-listing Maps pack for the core Edinburgh searches. That looks fair given the review count is higher than every rival shown.
2Branded AI answers4 / 48 of 8 branded prompts across ChatGPT, Gemini, Perplexity and Claude mention Harbour & Key, and every answer is accurate and positive. ChatGPT, Perplexity and Claude cite example-property.co.uk directly.This is already the aim: every tool names Harbour & Key correctly and positively when asked about it by name.
3Seller AI answers1 / 44 of 32 seller-intent answers name Harbour & Key: ChatGPT 1 of 8, Gemini 2 of 8, Perplexity 1 of 8 and Claude 0 of 8. Google's AI Overviews appeared on 8 of 12 searches and cited Harbour & Key on none.Named steadily when a seller asks something like “who should I sell my flat with”, and not only when asked about Harbour & Key directly.
4Searches that matter1 / 4Outside the top 30 on all 8 seller searches we tried: estate agents edinburgh, best estate agent edinburgh, solicitor estate agent edinburgh, fixed fee estate agent edinburgh, sell my house edinburgh, cheapest way to sell a house edinburgh, estate agents glasgow and best estate agent glasgow. The only rankings on these terms are national informational searches at 27, 28 and 32.A page one place for at least the fixed fee and solicitor estate agent searches, the terms closest to what Harbour & Key offers.
5Search breadth2 / 41,181 ranked keywords and an estimated 5,783 organic visits a month. 356 of the top 500 come from /property/ listing pages that expire when the property sells, and 31 non-listing keywords sit on page 2 (positions 11 to 20).Visibility that builds on lasting pages (guides, area hubs, /fees/), which stay when a property sells.
6Mobile speed1 / 4The homepage scores 44/100 on mobile, and the main content shows at 16.7 s against a 2.5 s target, with 746 KiB of unused JavaScript and a 4.0 MB page. Desktop is fine at 82/100 and 2.0 s, which suggests the build is the cause.Main content showing in under 2.5 s on a phone, matching what the desktop build already reaches.
7Headings and titles1 / 442 of 53 core pages have no H1 heading, including /sell/, /buy/, /fees/, /valuation/ and every guide step. 52 of 53 have no meta description, and 8 builder partnership pages (/barratt/, /taylor-wimpey/ and similar) have an empty title tag.Every core page with a clear H1, a written meta description and a real title, which Google and the AI tools need to quote a page correctly.
8Structured data1 / 4Across a 231-page crawl there is no RealEstateAgent or LocalBusiness schema, no FAQPage and no Offer or price markup. Only the theme's defaults (Organization, WebPage, Article, Breadcrumb) are present.The £600 fixed fee, the 4.6 rating from 662 reviews and the solicitor-plus-agent model written in schema, as well as in the page text.
9Content library3 / 4Around 550 posts back to 2012, guides of 1,000 to 1,900 words and no thin core pages. Some rankings already: “best areas to live in edinburgh” at #17 (590 a month), the LIFT scheme guide at #16 (720 a month), and LBTT/ADS guides at 27 to 31 (720 and 590 a month). The gaps: 60 posts have no meta description, legacy posts carry up to 14 H1s, and nothing is set up as a direct answer.The library updated with direct-answer capsules and clean headings, so existing rankings could turn into AI citations.
10Backlinks3 / 4382 referring domains (358 to the main site), authority rank 247 and a moderate spam score of 14. espc.com is the strongest linker. 36 broken inbound links waste link value, including 7 that point at a valuation page URL that returns an error.The broken links fixed (about an afternoon's work) and the profile kept clean. This looks adequate for a business this size, and it is probably not the lever for AI visibility.
11Reviews4 / 4662 Google reviews at 4.6, the most of any Edinburgh agent in this audit (ESPC 648, VMH 321, Neilsons 194). Profiles on allagents.co.uk and Trustpilot exist, and ChatGPT already cites them on branded questions.This is already the aim: what Maps and the AI tools need is in place and working.
AI visibility

The same picture in six smaller checks

CheckScoreWhat we saw
Entity clarity1 / 4The tools may struggle to fix what Harbour & Key is and where it works from a single South Gyle profile and thin core pages.
Answer formatting1 / 4No content is set up as a direct answer: no 40 to 60 word capsules and no question-style headings on the money pages.
Structured data1 / 4No RealEstateAgent, LocalBusiness, FAQPage or Offer schema across the 231 pages crawled.
Third-party mentions2 / 4allagents and Trustpilot profiles exist and get cited on branded questions, but Harbour & Key is mostly missing from the listicles and aggregators the tools lean on for seller questions.
AI crawler access4 / 4GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot all get a 200 response, and robots.txt is open. Nothing stops the tools reading the site.
Seller topic coverage2 / 4The seller guides are largely written (guides, LBTT/ADS content), but they are not shaped around the questions sellers put to AI tools.

These six checks do not feed the overall score. They show why Harbour & Key is named on branded questions and rarely on seller questions. Nothing blocks the AI tools from reading the site, and there may be little set up for them to lift.

Grading method

How each area is scored

ScoreGradeMeaning
4AStrong, worth protecting
3BGood, with some gaps
2CMixed
0 to 1DWhere the plan starts

The overall score adds up the eleven area scores with no weighting: 1 + 4 + 1 + 1 + 2 + 1 + 1 + 1 + 3 + 3 + 4 = 22 of 44.

Summary

What the scores point to

Rows 2 and 11 score full marks. Harbour & Key has 662 Google reviews at 4.6, the most of any Edinburgh agent we measured, and all 8 branded prompts named it accurately and positively. The content library scores 3 of 4, with about 550 articles back to 2012.

Rows 1, 3, 4, 6, 7 and 8 score 1 of 4. Row 7 is mostly missing H1s and meta descriptions, and row 8 is missing schema. Both look like fixes that could take days.

Rows 5 and 10 sit in the middle. 356 of the top 500 ranked keywords come from /property/ listing pages that expire when a property sells, so that visibility may not build up the way lasting pages could. The 36 broken inbound links could be fixed in about an afternoon.

From what we can see, the reviews, the articles and the £600 fixed fee are already in place. They may not yet be where Google and the AI tools read, which could be why six areas sit at the bottom while reviews and branded AI answers sit at the top. The plan starts from 22 of 44.

Method and limits

We scored each area against the grading above, using live checks on 8 July 2026. AI answers vary between runs, so a single run is a snapshot.

Real client work. Names and locations changed.