Nobody reads your homepage in order

You wrote your homepage as a document. Headline, then a paragraph about what the thing does, then some proof, then pricing, then the part where you ask for the signup. The argument is built in order, one piece resting on the one before it, because that is how writing works and you are a competent writer. Then somebody arrives from a search result, spends four seconds on the page, and leaves. They did not read your argument. They did not read anything. They scanned, in a rough diagonal, for evidence that they were in the right place, and they did not find it fast enough.

This is not an attention-span complaint, and the fix is not to make everything shorter. It is that a page has an order and a visit does not. You built a sequence; they are running a filter. Until those two things line up, every other improvement you make — better copy, better offer, more traffic pointed at it — is landing on a page that most people never actually reach the middle of.

The three questions people arrive with

A first-time visitor is holding three questions, and they resolve all of them or leave: what is this, is it for someone like me, and what happens if I click. That is the whole filter. It runs in a few seconds, mostly on whatever is visible without scrolling, and it runs before any of your persuasion has a chance to start.

So go look at your own first screen and mark which of the three it answers. Most answer one and a half. The usual pattern is a headline that states an aspiration — grow faster, do more with less — followed by a subheading that explains the mechanism, and nowhere in the visible area does it say who this is for. That part exists; it is in a section called "Who it's for," two thirds of the way down, which is precisely where the people who needed it had already gone. The aspiration headline has a second problem too: it is usually written in the category's polished vocabulary rather than the reader's, which is the same gap we went into in the words your customers actually use.

people rarely leave because they are not interested. they leave because they could not tell.

The third question gets mishandled in a quieter way. The button assumes a decision the visitor has not made yet. Start your free trial is a fine thing to ask of someone who has understood the product and is weighing it; it is a strange thing to ask of someone who is still working out what category you are in. When the only action on the screen is a commitment, the person who is not ready has nothing to do except leave, and you never find out that they were close.

What it is worth, on paper

Invent numbers, because the shape is the point. Say a thousand people land on the page in a month, and a little over half never scroll past the first screen — a fairly ordinary split, not a disaster. Of the four hundred and fifty who do go further, six percent sign up: twenty-seven signups. Now change nothing except the first screen — same product, same traffic, same offer, but the visible area answers all three questions instead of one and a half — and suppose that moves the share who scroll on from forty-five percent to sixty. Six hundred people go deeper, the same six percent convert, and you have thirty-six signups instead of twenty-seven. Nine more, from rewriting a screen's worth of words, and unlike nine more from buying traffic, this kind does not come with a monthly bill attached. It is the cheapest thing on your list precisely because you already paid for the visits.

Write for the scan, then for the reader

Here is the practical version. Your subheadings are the page in miniature, so read them on their own, top to bottom, ignoring everything between. Do they tell a coherent story to somebody who reads nothing else? If they say Features, Benefits, Integrations, FAQ, then your scanner learns nothing at all — those are filing labels, not sentences. Headings that make claims, in plain words, do the work of the whole page for the majority of visitors who will only ever see them.

And now the tradeoff, stated plainly, because writing for the scan has a real cost. Blunt scannable statements can hollow out a page for the careful reader, and the careful reader is usually the person closest to buying — they are reading slowly because they are seriously considering it. So do not trade one for the other. Put the argument in the headings and the opening sentence of each block, where the scanner will get it, and put the nuance, the caveats, the comparisons and the actual number you charge in the paragraphs underneath, where the person who has slowed down will find them. Two audiences, one page, different depths.

This is also where AI drafting quietly works against you, and it is worth knowing the failure in advance. When we draft a landing page or an ad, the draft arrives in document order — a clean opening, a developing middle, a conclusion — because that is what fluent writing looks like. It is the correct shape for an essay and the wrong shape for a page nobody reads in sequence. Which is why the draft waits for your approval rather than going live on its own: the edit that matters most is almost never stylistic. It is structural. Move the sentence that says who this is for up to the top. Delete the warm-up paragraph. Turn the label into a claim. And if the page is the destination for an ad, the first screen has one more job on top of all this — matching the promise that got them to click, which is its own failure mode we covered in when your ad and your page disagree.

There is a test for all of this that takes five seconds and needs no tooling. Show your first screen to someone who does not know what you do, count to five, take it away, and ask them two questions: what do we sell, and who is it for. Whatever they say back is your homepage. Not the one you wrote — the one that exists.