I have written a blog post every day for a hundred and ten days. This morning I sat down and asked what all of it had actually done.
The answer took the AI about twenty minutes to pull. Over the last ninety days, a page of mine showed up in someone's Google search results 4,244 times. Thirty-seven of those turned into a click. It is the first time I have sat down to check my SEO with AI properly instead of guessing at it.
I did not hate the number. I have known for a while that this is a slow build. What I did not expect was where the problem actually sits.
How I did it
No new tool. I opened my analytics in Chrome and pointed the AI at it. Two passes. First Google Analytics, for what got read and where the readers came from. Then Google Search Console, for what Google shows me for, and where I rank.
One snag worth knowing before you try this. My Search Console property was not verified on the Google account I was signed into, so the first pull came back empty. Ten minutes to verify the domain and it worked. If you do this and the numbers look impossibly thin, check that first.
The instruction was blunt. Pull every search term each page ranks for. Every page sitting on page two. Every query with a lot of impressions and almost no clicks. Put it in a table. Then tell me, in plain English, the five things I should do about it.
Seen, not chosen
My posts are not failing to rank. They are failing to get clicked.
The two posts about building an AI financial controller pulled more than 1,500 and nearly 1,400 impressions each over ninety days. Both sit on the first page of Google, around positions seven and eight. Between them they earned five clicks. A related pillar page ranks at position 5.6, near the top of page one, on 171 impressions, and got zero clicks. Not one.
That is not a ranking problem. Google is putting my work in front of people. They read the title, they read the two grey lines underneath, and they move on. The posts are seen and not chosen. A hundred and ten days of writing built for search, getting shown and skipped, because the title and the little description under it do not make anyone want the click.
There is a humbling footnote in the data too. Search Console only names about three percent of the searches that reach me. The rest are rare phrases it hides for privacy. So most of what people actually type to find me, I cannot even see.
The one page people click
One page breaks the pattern. It is the page about my son.
Finn died three years ago. The page that carries his story, and the fund we set up in his name, ranks for people searching for what happened to him, and it is the only page on the whole site that people reliably click. Just under five percent of the time, far better than anything I have built on purpose.
I do not have a tidy lesson to hang on that. The most human page is the one people choose, and the hundred I optimised for keywords are the ones they pass. I am still sitting with it.
The fade
The AI caught one more thing I would have missed on my own. A new site gets a short lift from Google when it first gets indexed. Mine started showing up in March. Impressions spiked in May, then dropped for two straight months. My average position has slid from about eight in May to seventeen now. The early lift is wearing off, and the long tail that carried it is drying up. If I want the line to keep climbing, the writing alone will not do it.
What I am changing this week
Three things, all cheap, none of them more writing.
Rewrite the titles and the descriptions on the page-one posts that get no clicks. The ranking is already there. The words that earn the click are not.
Push the handful of posts sitting on page two, at positions nine to fifteen. One of them ranks for "yahoo finance vs bloomberg" and is a single nudge off page one.
Tag every link I share on LinkedIn or anywhere else so my analytics can tell me where readers come from. Right now two thirds of my traffic shows up as "direct," which means the tool cannot tell me where those readers came from. I cannot fix what I cannot see.
I wrote back on Day 79 about auditing this site for search. That pass told me the site was technically healthy. Healthy and ranking is not the same as read. The check I had never run was twenty minutes of work, and I had put it off for a hundred and ten days.
Monthly Revenues $11,000 | Clients 2 | Prospects 1 outbound live, Meta and WhatsApp still down
Day 111 of 365.