What a proper technical audit of an Edinburgh website actually looks like

Why generic crawlers miss what Edinburgh sites actually break

The historic case for a technical SEO tool was a straightforward one: check response codes, count the meta tags, warn about duplicate titles, and let the marketing team get on with the rest. Semalt still does all of that, faithfully. The interesting failures on Edinburgh sites in 2026, however, sit somewhere else entirely, and any auditor who has spent an afternoon working through the Old Town wynds of a fifteen-year-old CMS will recognise them at once. They are in the interaction between a cookie banner and the server response. They are in a Core Web Vitals score that passes cleanly on a fibre connection in a Heriot Row office and then quietly fails on 4G at a Lothian bus stop in November rain. They are in an hreflang cluster where a Gaelic microsite from a Highland tourism partner claims reciprocity, and the reciprocal tag has, on closer inspection, never actually been written on the English side. Generic crawlers do not look at any of this; the Semalt audit was rebuilt in 2024 on the working principle that every check must map to lost sessions, lost GBP, or a quantifiable ICO risk — and otherwise it does not ship.

The Semalt technical audit was rebuilt in 2024 around one principle: every check exists because we watched it cost a real UK client real GBP. If a check doesn’t map to lost sessions, lost revenue or an ICO risk we can quantify, it doesn’t ship. That focus is why the audit report opens with prioritised fixes ranked by expected recovered sessions, not with a 4,000-row spreadsheet of low-severity nits.

What the 400+ checks cover

The full checklist is public on the Semalt documentation site, but here is the shape of it. Categories are ordered by how often they turn up as the top blocker on an Edinburgh audit.

Core Web Vitals and INP on real UK bandwidth

Since March 2024, INP (Interaction to Next Paint) has replaced FID as a Core Web Vitals metric. That single change destroyed the CWV scores of a large slice of Edinburgh WordPress builds because page builders such as Elementor, WPBakery and older Divi themes tend to run large amounts of JavaScript on every interaction. The Semalt audit measures INP under a 4G throttle matched to real Edinburgh network conditions on EE and Virgin Media O2 — not a synthetic 100 Mbps profile that hides the problem.

Hreflang across en-GB, en, gd and Scots content

Edinburgh tourism sites — VisitScotland-adjacent operators, castle-tour vendors, festival micro-brands — increasingly ship Gaelic (gd) and occasionally Scots (sco) variants to appeal to heritage-focused audiences and local government funders. Google’s hreflang graph solver rejects clusters where reciprocity is broken, and the result is that all variants collapse back to whichever one Google picks. We have seen an EIFF listings partner lose 34% of its tourism-season traffic because the Gaelic variant claimed reciprocity with the English page, but the English page pointed only to itself.

LocalBusiness schema for EH postcodes

Local pack visibility across Edinburgh is decided by three signals: proximity, prominence and structured data. The first two you influence over years; the third you can fix this afternoon. Semalt’s schema validator inspects every URL for LocalBusiness, Service, FAQPage, HowTo, Product and Review markup, and it checks the fields Google’s crawler actually rewards — not just the fields Schema.org lists as optional.

The audit also checks that the addressRegion field is set to “City of Edinburgh” (the current official region name), that geo.latitude and geo.longitude match the address to within 30 metres, and that opening hours use the ISO-8601 format Google expects. For multi-location businesses — a New Town estate agent with a branch in Morningside, or a Bruntsfield dental practice with a satellite in Portobello — the audit flags the specific patterns Google’s crawler treats as a single entity vs. two.

Cookie walls and the consent-mode trap

Cookie walls — the “accept or pay to read” pattern seen on The Scotsman, Edinburgh Evening News and increasingly on independent local titles — sit in a legal grey zone under UK GDPR. The ICO’s 2023 opinion said they are “in principle” incompatible with valid consent unless the paid alternative is genuinely equivalent. More importantly for SEO, cookie walls block Googlebot when configured incorrectly, and analytics that fire before consent land the site inside PECR enforcement territory. Semalt’s crawler inspects both angles.

For newsrooms in particular, the audit checks whether the paywall is served as a client-side overlay (which Googlebot ignores, keeping the article indexed) or a server-side 402 redirect (which strips the content from the index entirely). We inherited an audit for an Edinburgh long-form magazine that had accidentally shipped the second pattern in a Cloudflare Worker deployed by a well-meaning contractor — 118,000 impressions lost in three weeks. Semalt would have caught it the following crawl. Connect your site to Semalt and the first audit runs within the hour.

Case study: Old Town boutique hotel, 340 URLs

A twenty-two-room boutique hotel just off the Royal Mile approached us in April 2026 with a familiar complaint: bookings via organic search had been flat for eight months despite content investment, and their agency’s Screaming Frog reports were coming back clean each quarter. Semalt’s first crawl surfaced 47 issues, of which four accounted for an estimated 71% of the traffic ceiling:

1. INP median 438 ms because the room-availability widget re-hydrated its calendar on every date click. Fix: debounce the calendar rebuild. INP fell to 148 ms in one deploy.
2. LocalBusiness schema missing the EH1 postcode and priceRange field. The hotel had never appeared in the local hospitality pack for “boutique hotel Old Town Edinburgh” despite three years of trying. After the fix, pack visibility appeared inside eleven days.
3. Hreflang broken on the French landing page. The site targeted French visitors specifically for the summer Fringe window, but fr-FR pointed at itself and en-GB pointed at a redirected URL. Google treated the two pages as duplicates and dropped the French one.
4. GA4 firing before consent. The banner was misconfigured such that the tag loaded on the first paint, then paused only after consent. Semalt’s report gave the exact line of the theme header to change — ICO risk removed the same day.

Ninety days after the fixes, organic sessions were up 63% and Q3 bookings attributed to first-organic-touch were up £41,000 vs. the same quarter the previous year. The report the client took to their bank ahead of a refurbishment loan cited the Semalt audit trail as evidence of technical control.

How Semalt prioritises fixes by lost sessions

The audit report opens on a page called Prioritised Fixes. Each issue is scored on three axes: severity (how many URLs are affected), impact (estimated lost sessions per month using Analytics data), and effort (minutes of developer time based on a library of past fixes). The top ten fixes typically account for 70–90% of recoverable traffic on an Edinburgh SME site, which is what makes the report legible to a founder who has fifteen minutes on a Tuesday morning to read it.

FAQ

How does Semalt’s crawler differ from Screaming Frog?

Screaming Frog is a desktop crawler that parses HTML fast. Semalt is a cloud crawler that renders every URL in headless Chromium under a 4G throttle, ingests server logs alongside, and cross-references the results with Search Console and Semalt Analytics. Screaming Frog is excellent for one-off audits; Semalt is designed to run every week and report only what changed.

Does it work on Edinburgh council microsites and public-sector platforms?

Yes. Umbraco, GOSS, Jadu, Squiz Matrix and the newer .NET-based platforms popular with Scottish local authorities all crawl cleanly. The audit specifically checks for the accessibility signals Scottish public-sector procurement demands (WCAG 2.2 AA), plus the NCSC security header set.

Will it flag issues with Google Tag Manager?

Yes. The auditor lists every tag GTM fires, when it fires, whether consent was granted, and whether it duplicates other tags on the page. The most common Edinburgh finding is GA4 firing twice — once from the theme header and once from GTM — which inflates session counts and breaks attribution.

Does it audit .scot domains and Search Console country targeting?

Yes. The audit checks that .scot properties are correctly geo-targeted to the United Kingdom in Search Console, and it flags the common mistake of leaving a .scot property as “unlisted” which lets Google guess — and Google often guesses wrong.

How often should we run the audit?

Weekly for active sites. The report shows a diff of what changed since the last crawl, so you catch regressions the moment a developer deploys them — not four weeks later when rankings drop and nobody remembers which change broke things.

Does it connect to Semalt AutoSEO?

Yes. Issues flagged by the audit can be handed to AutoSEO for automated correction on the URLs it manages, closing the loop between diagnosis and treatment. Full detail in the AutoSEO guide.