What a technical SEO check actually finds
Not keywords. Mostly pages Google cannot see, pages it sees twice, and pages that are too slow on a phone to matter. The findings that come up again and again on small Danish business sites.
When people hear SEO they think of keywords and content. A technical check is the part underneath: whether search engines and AI assistants can fetch the site, understand it and trust which version of a page is the real one. On small business sites the same handful of problems turn up again and again.
Pages Google cannot see#
A robots.txt or a "noindex" left over from when the site was built on a staging domain.
A bot filter or firewall that blocks crawlers along with the spam, so Google and the AI assistants get an error page.
Content that only appears after JavaScript runs, on a site where the server sends an almost empty page.
A sitemap that is missing, out of date, or lists pages that redirect.
Search Console usually tells you about the first and last of these, if somebody looks. I wrote about what the crawlers actually fetch in three days of crawler logs.
Pages Google sees twice#
The site answers on both www and the bare domain, or on both http and https, without redirecting one to the other.
The same product or article reachable under several URLs, with no canonical tag saying which one counts.
Language versions without hreflang, so the Danish and English pages compete for the same query.
Duplicates do not get a site punished, but they split the signals between copies, and Google picks which one to show, not you.
Pages too slow to matter#
Most Danish traffic for local businesses is on phones. The usual causes are oversized images, a theme or page builder that loads everything on every page, and third-party scripts: chat widgets, tracking, cookie banners. Core Web Vitals in Search Console show whether real visitors are affected, which matters more than a single test score.
Snippets that say the wrong thing#
Old titles and descriptions stay in Google's results long after the page changed. A price that is no longer valid, an old address, a service you no longer offer. Updating the page and asking for it to be crawled again fixes it faster than waiting.
Structured data and the local profile#
For a local business the Google Business Profile and the organisation data on the site should agree: name, address or service area, phone, opening hours, website. When they disagree, both Google and the AI assistants have to guess.
What a check is not#
A technical check does not write content and does not promise rankings. It removes the reasons a good page would not be found. The content itself is a separate job, and one I point people elsewhere for.
What the check covers: technical SEO and GEO.
Was this useful?
More notes
What three days of crawler logs said after I shipped an llms.txt
I put the GEO plumbing on micci.dk — an llms.txt, deliberate robots rules, and one log line per crawler — and then read the log. The assistants fetch robots.txt and sitemap.xml far more than they fetch pages, and a user-agent turns out to be a claim rather than an identity.
Taking over a Laravel app after the developer left
The app works, the person who built it is gone, and nobody knows what happens if you touch it. What I look at in the first week, in the order that stops things getting worse.