Four of my five keywords died on search intent

I was picking topics to bring readers to a blog. I measured monthly search volume, looked at competition, and took the five biggest.

An hour later I threw four of them away.

Five picked without reading search intent

Here is what I had, chosen without looking at search intent at all:

Keyword	Monthly searches	The post I planned
timer	102,400	Three weeks of living by a 25 minute timer
study planner	20,740	Moving a paper planner into an app
timetable	8,270	Why the timetable you wrote is not followed
ritual	2,180	Ritual and routine are different words
three year diary	590	Counting days versus accumulating them

On the numbers this is a good list. One keyword at six figures, the rest small enough that a personal blog could realistically rank, and no reason visible in the data to question the search intent behind any of them.

There is one thing I had not done. I never checked what the people typing those words wanted to see.

I opened the pages that actually rank

Downloading a search results page over plain HTTP gets you ads and navigation chrome, because the results are rendered by JavaScript. So I drove a real browser, ran each search, opened the top organic results, and read the article bodies.

The result:

Keyword	What the ranking pages are actually about
timer	Price comparison, dollar-store kitchen gadgets, a 20 second dance challenge
timetable	Train timetables, intercity bus terminal departures
ritual	Sisley shampoo scalp care, film reviews
study planner	Cram school ads, price comparison, sponsored planner reviews

None of the four has anything to do with the posts I intended to write.

Someone typing “timer” wants to buy a kitchen timer. “Timetable” is what you type to find a train. And in Korean, “ritual” has largely been claimed by cosmetics marketing.

Volume tells you the size, not the reason

This is the part I had wrong.

A keyword tool tells you how many people typed a string. It does not tell you why they typed it. For an article to get read, the second one has to match.

You can rank first for a six figure keyword and still get nothing, if those hundred thousand people came to buy a timer. In that case the volume number was not telling me how many people would read the post. It was telling me how many people could bounce off it.

Ambiguous words are where this bites

Three of the four died the same way: the word has more than one meaning, and most of the volume belonged to a meaning I had not considered.

The same thing happened with a Korean phrase I had used for “streak”. I wanted to write about habit streaks. The ranking pages were about navigation apps recording continuous driving, baseball records, and a phone’s burst photo mode.

⚠️ You cannot filter this by thinking harder. In my head “ritual” meant a morning routine, and nothing in the volume data hinted otherwise. Until you open the ranking pages, you will keep believing your own definition is the world’s definition.

So I changed the order

It used to be:

measure volume -> pick topic -> write -> run SEO check

Now there is a step in the middle:

measure volume -> read the ranking pages -> pick topic -> write -> run SEO check

And I made the checker enforce it. A post aimed at acquisition does not pass unless a research file exists for its primary keyword.

Writing the step into a document was not enough. I wrote eight posts the same day without doing it once, and not because the instruction was missing. It was because skipping it had no consequence.

Even the survivor needs a different angle

Study planner was the one that stayed related, but the pages ranking for it are cram school ads and sponsored reviews. An informational post there is sitting between advertisements.

I read that as a signal to change the angle rather than drop the keyword. If people are in the middle of choosing a planner, the thing to write is a comparison, not a how-to.

What this does not fix

Reading the ranking pages tells you what a keyword currently means. It does not tell you whether you can outrank the pages already there, and it does not tell you whether the people who do land on your post will care about anything you sell.

Those are separate questions, and I have not answered either. What this step removes is one specific way of wasting a week: writing a careful post about a word that means something else to everyone but you.

It is also cheap. Five pages per keyword, about thirty seconds, cached for the day. Compared with writing a post nobody wanted, that is nothing.

Scope

Measured on 2026-08-15 against Korean search results. Volume figures come from Naver’s search advertising API, and for each keyword I opened up to five ranking pages with Playwright 1.60 on Node 24 and read their bodies.

I did not look at Google. Search results vary by region and change over time, so the same keywords measured on another day may come back different. This is also a sample of five keywords, not a study.

Whether the new ordering actually increases traffic, I do not know yet. I have only just changed it.