March, a rented desk in Tallinn, and a spreadsheet with one row highlighted in yellow. That row held a broad phrase with four thousand searches a month, and I had convinced myself it was the only thing standing between this site and an actual audience. I gave it one weekend and told myself the rest would follow.
The month the numbers looked right
I published the piece on a Tuesday and read the ranking report every morning like a departures board. Six weeks in it sat at position eight, then position four, and the sessions graph climbed at a slope I had never seen on this site. For about ten days I was certain I had found the mechanic.
What I did not check was what happened after the click. Median time on page was twenty-two seconds, scroll depth stopped at the second heading, and the comments field stayed empty. The traffic was real, the readers were not.
A friend who runs analytics for a living looked at the numbers over coffee and used the word that stung: mismatch. The page answered a question nobody had asked in the way I had written it, and the graph had been flattering me for weeks.
That was the first time a figure on the site told me something the chart could not. Sessions rose because the phrase matched a habit people already had, not because anyone wanted four thousand words on the subject. I kept the row in the plan for two more months before admitting the difference.
What the phrase was actually about
I opened the top five results with a notebook beside me and read them the way a visitor would. Two were free tools, one was a forum thread with the fix in the fourth reply, and one was a short definition that finished in four hundred words. None of them were essays, and mine ran four thousand words with a story at the top.
The searchers were mid-task. They wanted a field cleared, not a philosophy, and my introduction asked them to care before it solved anything. I had written for the reader I wanted rather than the one the query produced.
Intent shows up in small ways. Someone scanning for a word they already know will not read three paragraphs to find it, and my page made them do exactly that. Once I moved the actual answer into the first forty words of a later rewrite, the same piece held readers past a minute with nothing changed except sentence order.
My notebook from that week holds a line I still use when I plan a piece: who is standing at the other end of this sentence. If the answer is someone with a problem and eleven minutes, the story has to earn its place after the fix instead of before it.
Deleting the row
Twelve months after that first weekend I deleted the row from the spreadsheet. Nothing dramatic happened: the post stayed, traffic settled at a third of its peak, and I stopped refreshing the report each morning. The point of deleting it was to stop steering the calendar toward a number that never paid back.
In the same session I added two narrower phrases, each with under a hundred searches a month, and wrote one piece for each. They took an afternoon instead of a weekend, and between them they produced six emails from people who wanted a straight answer with a price in it.
The narrower work also changed how a week gets planned. Instead of one long piece that has to carry the whole month, the calendar now holds three short ones that each stand alone when a season turns busy. Two hours on a bus is enough for one of them, so a trip with bad signal no longer stalls the schedule.
Neither phrase will move a graph the way the broad one did for ten days in April. Both brought readers who finished the page and came back a month later, which is the only figure I write into a plan now.
What happens before writing now
Every idea gets opened against the current results page before it earns an outline. I read the top five and ask whether mine would be the most specific answer there, and if page one is full of tools, agency pages and old threads, I pick a narrower phrase instead. That takes ten minutes and has saved me weekends.
The second test is the intent bucket: problem, comparison or decision. A page written for a problem query that ranks for a decision query converts badly, and the reverse is worse, so the bucket goes into the top of the outline while the idea is still fresh.
There is one exception worth naming. If a broad phrase genuinely describes what the site is about, I will write for it, but only after the narrow versions exist and only as a piece that answers a real question on the first screen. Reach is something a page earns once, not something a volume figure promises it.
What replaced the deleted row is a habit rather than a tool. Before the first line gets written, I ask who is standing at the other end of the sentence and what they need in the next forty words.