Two versions of the same page. One opened with a broad promise about learning marketing. The other opened with a refusal: this is a notebook, not an agency and not a course. I wanted to know whether a sentence that names who the work is not for would change who stays, not whether it would win a beauty contest.
The idea underneath is in positioning is a refusal. The questions I used to write the sentence are in positioning questions. This page is the test, including the part that looks like a loss if you only watch the first screen.
The refusing sentence lost more of the first screen, and the people who stayed scrolled further.
What I believed
A headline that names who the work is not for will lose some readers immediately and keep the right ones longer than a broad promise. The broad promise should feel safer and teach less. The refusal should feel ruder and sort the room.
I did not predict a conversion rate. There was no cart. The behavior I cared about was whether a person who continued reached the later parts of the page: the point of view, the first real argument. "Longer" meant further down the same page, not more minutes with a tab abandoned in the background.
What I held still
The page underneath the opening was the same. The subject was the same. I did not change the offer, because there was no offer beyond reading. I did not change the type, the order of sections, or the links. One sentence at the top was the change. If I had rewritten the whole hero, I would have learned nothing I could repeat. That rule is the experiment note.
The audiences were people who had some reason to open a page about learning marketing. I am not going to invent a sample size to make this sound like a laboratory. The comparison was repeated enough that the shape showed up more than once, and not enough that I would bet a budget on the exact gap. The documented result is the shape, not a false decimal.
| Broad promise | Refusing sentence | |
|---|---|---|
| Opening | Learning marketing, said widely | A notebook, not an agency or a course |
| First screen | More people stayed briefly | More people left immediately |
| Later on the page | The ones who stayed tended to stall in the same place | The ones who stayed tended to go further |
| What it felt like to publish | Safer | Easier to dislike |
| What I would repeat | The control, as a control | The refusal |
Safer, and forgettable
The broad version felt safer to publish. It also gave a reader no reason to sort themselves. They stayed long enough to be polite and left around the same paragraph every time. Polite reading is not interest. It is the absence of a reason to leave yet. When the reason arrived — the page still had not said who it was for — they left together, which is a kind of evidence. The sentence had delayed the decision instead of making it.
I noticed myself wanting the broad version to win. It was less likely to annoy a friend. That desire is not a result. It is loss aversion wearing a social coat. Killing a soft sentence feels like turning a reader away at the door. The test was whether the door should have a sign. The sign worked. My discomfort was not a metric.
Easier to dislike
The refusing version was easier to dislike. That was the point. The readers who continued reached the philosophy and the first essay more often. I would repeat the refusal, and I would not hide it under a softer line to recover the people it was written to lose.
Dislike from the wrong reader is a successful sort. Dislike from the right reader would have been a failure of tone or truth. I did not get a clean way to label every person. What I could see was the path: leave early, or go further. The early leavers were the cost I had agreed to pay. The further reading was the thing the broad headline was not buying, even though it bought a few extra seconds.
If I had optimized the first screen alone, I would have shipped the broad line and called the refusal a failed test. That is how positioning dies in analytics. The first number is the one the tool shows. The second number is the one you have to care about before you open the tool. I wrote the second number into the hypothesis so I could not quietly drop it.
What this does not prove
It does not prove that every brand should open with a refusal. A refusal only helps if there is a real boundary. A fake boundary — "not for people who hate quality" — loses nobody and teaches nothing. The sentence has to be capable of stinging a plausible reader.
It does not prove a long-term effect. I did not follow the same people for a month. Further on one page is not a relationship. It is a better first sort. Relationship would be a later test: do they come back, and do they come back for the notes rather than for a vague sense that the site is "about marketing."
It does not prove the exact wording is sacred. Another refusal might sort as well. The lesson I am willing to keep is the shape. Lose the mismatch early. Keep the match long enough that they meet the actual point. Do not use the first-screen count as the only judge.
What I would run next
I would keep the refusing sentence and change one later paragraph, not the headline, to see whether the people who stay still stall somewhere specific. If they stall, the opening did its job and the middle is the next constraint. That is attention applied after the sort. Mixing the two fixes in one edit would make the next result unreadable.
I would also write down, in a sentence, a wrong reader I am willing to lose by name of situation, not by insult. "Someone who wants a done-for-you campaign." Then I would check whether the page still flatters them anywhere below the headline. A refusal at the top and a welcome in the third section is two experiments wearing one URL.
The broad line stays in the notes as the control, not as a backup I publish when I feel shy. Shyness is how the safer sentence returns. The documented result is enough for me to keep the refusal on purpose. It is not enough to pretend I measured a market. I measured a page. That is the size of the claim, and I would rather keep the claim that small than inflate it into a rule for other people's businesses.