Two Weeks of Search Console Data Falsified My SEO Playbook
I run lans.cloud. It is a collection of 100 free browser tools.
I started with a standard SEO playbook. I targeted long-tail phrases. I built deep topic clusters. I linked everything together. I waited for results.
After 14 days of data, the numbers proved my playbook wrong.
Here are the five lessons I learned:
- Split named versus hidden queries
Search Console hides many queries for privacy. This means you cannot see them. On my site, 47% of impressions came from hidden queries.
The hidden queries ranked 16 positions better than named ones. But they converted 2.5 times worse.
One page had 555 impressions at position 9 but zero clicks. I thought my title was bad. I rewrote it. Nothing changed.
The real issue? The named queries were at position 34. The hidden queries were at position 7. Google answered the users' questions directly in the search result. They did not need to click.
Do not rewrite a title just because clicks are low. Check if the search result already answers the question.
- Look at who owns page one
The playbook says deep clusters rank. My biggest cluster sat at position 37. My smallest cluster sat at position 7.
The difference was the competition.
My winning pages target intent that no brand owns. Page one is just a mix of small sites. My losing pages compete with giant brands or Google widgets. You cannot win those fights with content alone.
Before you build, search the phrase. If a major brand or a Google widget owns the top spot, do not build it. If you see a mix of small sites, build it.
- Beware of "variant explosion"
A page can have many different phrasing variants. This is not always good.
One of my pages had 60 variants but sat at position 62 with zero clicks. Another page had 40 variants and sat at position 7 with a 16% click rate.
High impressions from many phrases do not mean you are winning. They only amplify your actual rank. Check your position first.
- Decompose your aggregate trends
My site CTR went from 5.5% to 8.6% in two weeks. I did not celebrate.
The improvement came from one single page getting more impressions. One page held 77% of all my clicks. Most of my other pages had zero clicks.
If you only look at the total site numbers, you will see fake progress. Break your data down into individual pages before you make decisions.
- Specificity beats generic SEO
I tested keyword cannibalization. I had a generic "flip a coin" page and a specific "rigged coin flip" page.
Google ranked the specific page at position 11 and the generic page at position 83. The page with personality won the generic term. The page built only for SEO failed.
Summary of lessons: • Split named versus hidden queries. • Check the SERP competition before building. • Depth only works if the search results are open. • Never trust a total average without looking at the parts. • Phrase variety is an amplifier, not validation.
Source: https://dev.to/lans-cloud/two-weeks-of-search-console-data-falsified-my-seo-playbook-51if