The fifteen minutes that ruin it
You reread your post, it's fine — then you reword it "to make it better." Fifteen minutes later, it reads like nothing. The trap to avoid.
Sunday night. Your post is ready, fine, it says what it needs to. And then, that little voice: "I can do better." You reword the first sentence. Then the second, so they match. Then you smooth the paragraph. Fifteen minutes later, you have a clean, fluid text… that reads like nothing. You've just crossed the threshold where editing stops improving and starts undoing.
That threshold isn't where you'd think. The classic fear is publishing too raw — something unchecked, not quite right. The real danger is also on the other side: by dint of improving, each edit makes the text smoother and less yours. The generic isn't only what an AI produces. It's also what you make yourself, when you edit one time too many.
The question that stops the slide
One question, asked at every edit: is it wrong, or is it just not how I'd have said it?
If it's wrong — a false fact, last season's price, a sentence promising a service you don't offer, a hook whose follow-through betrays the tone — fix it. It's surgical, local, done in seconds: a word, a number, a line, and the rest holds.
If it's only "not how I'd have said it" — the sentence was fine, you'd just have turned it differently — stop. That edit is the first step of the slide. One more pass. Five in a row, and you're not reviewing anymore: you're rewriting, and rewriting downward.
Why improving to death flattens
Rewriting looks like fixing. It isn't.
When you reword by reflex, end of day, under the pressure of "it has to be perfect," you don't pull the text toward your voice. You pull it toward your defaults — and your defaults, tired and rushed, are closer to the average than the previous version was. You sand off the edges: the slightly blunt phrase, the too-precise detail, the angle that gently divided. Exactly what made the text recognizable. It's the generic everyone dreads — except here, you're the one reintroducing it.
What's worth your minutes
Your review minutes are scarce. Spend them where your judgment is worth gold: catching what's wrong, what's off-brand, what promises what you don't deliver. That, no one will catch for you. The rest — the turns of phrase you'd have worded differently — leave it. Not because a text is sacred, but because that quarter-hour of edits doesn't make it better: it makes it smoother, and so more nondescript.
A simple cue: if you fix a line here, a number there, you're on the right side. If you touch nearly every paragraph, you've tipped over — and the real question is no longer "is it good?" but "why does it feel not-me?" Usually the answer is one detail you set once in the settings, not fifty edits.
The text doesn't need you to rewrite it. It needs you to catch what's wrong — and to stop there. The fifteen minutes you were about to spend were the ones that would have made it a little more like all the others.