We had 21 SEO tasks. Search Console data closed 10 of them
The eight-step method we used to decide which SEO tasks to keep, and what it showed on our site.
We had 21 open SEO tasks and stopped to check which ones had evidence behind them. The method joins Google Search Console to the search results page, the deploy history, URL Inspection and first-touch attribution in our own database.
The short answer
- Of 21 tasks, 5 had already shipped, we closed 10, and we kept 5. Four of the five kept tasks have a smaller scope than before.
- Keyword tools did not decide. We used their volume and difficulty estimates only to choose which terms to check against our own query rows.
- Content pages brought few signups. About 2% of new users first arrived on a blog post, example page or benchmark page. One free directory listing brought 24%.
The eight steps
- Take property totals with no dimension and no filter. Do not add up page or query rows.
- Split disclosed queries into brand, diagnostic (
site:searches) and other. Report the undisclosed remainder as unknown. - Check each keyword tool claim against the page's own query rows.
- Search the term and read the first page of results to check intent.
- Record the live date and the crawl date for every changed page.
- Follow signups by date and landing page, counting only real accounts.
- Give each task one result: already shipped, keep, keep with a smaller scope, or close.
- Review on a fixed interval that matches your volume.
When it applies
- Volume sets the interval: with a few dozen clicks in 28 days, a weekly review mostly reads noise. We now review every 28 days.
- Small samples: the percentages in the post come from a sample of a few dozen new users. Use counts until each cell has enough people to mean something.
- What it cannot tell you: cause, what the undisclosed queries were (42% of clicks in our latest window), and whether the output helped the user.
The full post has the error each step prevents, our numbers for each step and the two closures that came from primary sources.