Message First

In our evals, AI-built landing pages kept opening the same way: a big headline that could sit on any product's page, something like "Work smarter, together", a grey line under it, two buttons. Then eleven tidy sections. The spacing is right, the contrast passes, and you still can't tell what the product does. There was nothing to lay out, because no one decided what the page was saying.

Before you choose a single section, write down who the page is for, what they use today, and what only this product does, because no layout can rescue a message any competitor could say.

QuaysideSwitchyardCrumb Ledger

QuaysideSwitchyardCrumb Ledger

Draft 1Draft 2Draft 3

Draft 1Draft 2Draft 3

Swap in Harborly

Quayside

The all-in-one platform for modern marinas.

Streamline operations and delight your customers.

The message, written first

For

Harbormasters at marinas under 200 berths (not in this draft)

Instead of

A paper berth chart and the phone (not in this draft)

Only

Matches each boat's length to a berth that fits (not in this draft)

So

Visiting boats book at 2am without calling (not in this draft)

Ticked lines made it into this draft.

• Can you picture it? NoNo boat, no berth, no person. A platform isn't a thing you can see.
• Could it turn out false? NoNothing here could be checked and found untrue.
• Could nobody else say it? NoEvery marina tool says this.

Pick a product and step through its three drafts. The first could sell anything. The last names a thing you can picture, a claim that could turn out false, and a difference the competitor can't copy by changing the logo. Swap in the competitor's name to test it.

Start from what they'd use instead

April Dunford's positioning process runs in a fixed order: the alternatives customers would use if you didn't exist, then what you have that those alternatives don't, the value that follows, who cares most about it, and last, the category. She explains why the order matters: starting anywhere else gave her "positioning that sounded good in the office, but it didn't work with customers because it wasn't differentiated."

The alternative is rarely another startup. It is a spreadsheet, a clipboard by the dock, a phone call, or doing nothing. Naming it tells you what the page has to beat, and it gives you the reader's words, because that's the thing they already know.

So before any section exists, write four lines:

• For: the people who care most, narrowly enough that someone could say "that's not me".
• Instead of: what they do today.
• Only: what this product does that the alternative and the nearest competitor can't.
• So: what changes for them, in a thing they could see happen.

The third line is the "only" line from the direction note, rewritten for a buyer. The headline usually comes straight out of the third and fourth.

Three tests for every line

Harry Dry asks three questions of any line of copy, as Demand Curve reports: can you visualize it, can you falsify it, and can nobody else say it? "If you have three no's, you've probably written a lot of rubbish."

Visualize means a reader can picture something: a boat, a berth, a loaf. "Streamline operations" has no picture. Falsify means the claim could be checked and found untrue. "Better insights" can't be wrong, so it says nothing. Nobody else is the competitor test: put the nearest competitor's name in front of the line. If it still reads true, it's their headline too.

Julian Shapiro's version of the same test is for the header alone: if a visitor reads only that text, will they know exactly what you sell? His examples of headers that fail are "Improve your workflow!" and "Supercharge your collaboration!".

The message picks the sections

Once the four lines exist, they answer most layout questions. The "only" line becomes the headline and decides what the first screen has to show. The "instead of" line becomes the comparison the page needs. Every claim needs proof next to it, and every doubt a buyer could have becomes a section, in the order they'd ask it.

Without them, sections get chosen from a template, and each one gets filled with whatever sounds plausible. That's how a page ends up with eleven equal sections of small grey text: none of them had a job.

How sure is this? Dunford, Dry and Shapiro are practitioners writing from client work and tests they ran, not from controlled studies. The time pressure behind it is measured: in Nielsen Norman Group's analysis of page visits, people are most likely to leave in the first 10 seconds, and a page has that long to say what it offers. A headline that needs the rest of the page to make sense spends those seconds on nothing.