Lists of SEO mistakes are easy to write and hard to trust, because nobody says how often each mistake actually happens. So we did the obvious thing and counted on our own site.
On 2026-10-03 we scanned every URL on attensira.com, 632 of them, against a checklist of the things that cost rankings and AI citations, as the first step of rebuilding the site. This post goes through what it found, most frequent first, with how to detect each mistake and how to fix it at the root.
What did a full scan of our own site find?
The table at the top of this page has the counts: seven mistakes, most frequent first, each as the number of URLs it touched out of 632. Links to pages we had decided to retire led, on 205 URLs; missing breadcrumb markup came last, on 2.
None of these is dramatic. Together they touched a large share of the site. And every one of them came from the same cause: pages written at different times, by different processes, without a shared template enforcing the basics. That is the real lesson, and we come back to it at the end.
Mistake 1: Internal links to redirected or retired pages
What it is. A page links to a URL that redirects elsewhere, or no longer exists. Every redirect wastes a hop for crawlers and visitors; every dead link wastes the link entirely.
How it happens. A site consolidates pages, renames a section or retires old posts, and nobody updates the hundreds of links pointing at the old URLs. Ours came from a site architecture review that decided to merge or remove a large number of overlapping pages; the links had not caught up yet.
How to find it. Crawl the site and list every internal link whose target returns a redirect or an error.
How to fix it. Point each link at the final destination. Better, keep a single list of retired URLs and their owners, and have your build fail when a page links to one. That is what we now do: a test fails if any page links to a deleted URL.
Mistake 2: Meta descriptions that are too long, too short or missing
What it is. The description Google may show under your title in results. Google's starter guide says the snippet "may be sourced from the contents of the meta description tag". Too long and it is cut off; too short and it says nothing; missing and Google picks text from the page, which may be a cookie notice.
How to find it. Crawl and check length. Aim for roughly 120 to 160 characters that state what the page answers.
How to fix it. Write descriptions from the page's own first answer. If the page opens with a clear 40 to 60 word answer, the description is that answer, shortened.
Mistake 3: Titles that are too long
What it is. Titles cut off in results lose their ending, which is often the most specific part. Ours were mostly long because the brand name was appended to an already long headline.
How to find it. Crawl and check title length. Around 60 characters is a practical limit for most results layouts.
How to fix it. Lead with the query the page answers, keep the brand once at the end, and cut adjectives. "How to Calculate Share of Voice" beats "The Complete, Definitive Guide to Calculating Share of Voice in 2026 and Beyond".
Mistake 4: No direct answer near the top
What it is. A page that exists to answer a question but takes three paragraphs of introduction before it does.
Why it costs more now. Google says a page needs to be indexed and eligible to be shown with a snippet to appear in its AI features. AI engines and snippets both lift passages, and a page whose answer is buried rarely has the passage they choose. A searcher scanning ten results might scroll to find the answer; an engine writing one answer picks another page.
How to fix it. Put a 40 to 60 word answer directly under the main heading, written to stand alone. See answer blocks.
