Fix the page before you buy more traffic

There are two ways to double your signups this quarter. Double the traffic, or double the share of visitors who sign up. The arithmetic is identical; the price could not be more different. The first one you rent — it costs money every month, and the month you stop paying it stops working. The second one you own — you do the work once, and every visitor who arrives afterward, from every source, for as long as the page exists, walks into the better version. Almost everyone buys the rented one first.

There is a reason for that, and it is not stupidity. Traffic is purchasable on a Tuesday afternoon. You put in a card, pick an audience, set a budget, and by Wednesday the line on the chart is higher. Improving a page is vague, slow, and has no button. When one option is achievable this afternoon and the other takes a fortnight of thinking, the afternoon wins — not because anyone weighed it and decided, but because the afternoon is the one with an obvious first step.

Put invented but plausible numbers on it. You get 2,000 visitors a month and 2 percent of them sign up: 40 signups. Plan A, buy 2,000 more visitors at $1.50 a click — that is $3,000 a month, every month, and cold paid traffic almost always converts worse than the people already finding you, so call it 1.2 percent and 24 extra signups. Plan B, move the page from 2 percent to 3.5 percent: 70 signups from the same 2,000 visitors, no monthly bill, and the improvement quietly applies to Plan A's traffic too if you ever run it. Plan B wins in this example not because pages are magic but because a rate multiplies everything downstream of it, indefinitely, while a budget only buys the month you paid for.

traffic is rented. the page is owned.

What fixing the page usually means

Not a redesign. Almost never a redesign. In most cases the page is not ugly, it is vague — it describes a category instead of making a promise. A stranger arrives, reads the first screen, and cannot answer three questions: who this is for, what it does for them, and what happens if they click. Write those three answers in plain sentences, put them above everything else, and you have done most of the available work. The rest of the page can stay exactly as it is.

The second most common problem is order. Pages routinely ask for the meeting before they have earned any interest, or ask for interest before they have answered the question the reader actually arrived with. That is the same instinct behind [leaving the price off the page](/blog/put-the-price-on-the-page): a hope that the visitor will hand over their attention first and get their answers later. They will not. They read in the order their own questions arrive, and a page that refuses to follow that order feels evasive even when every individual sentence on it is true.

The obvious objection is that you cannot know what to change without testing, and at your traffic level a proper test would take months — which is true, and [less of an obstacle than it sounds](/blog/too-small-to-a-b-test). The evidence is already sitting around you. Read your own replies to the last twenty people who asked what you do; the sentence you keep typing by hand is your headline, and it is better than the one on the page because you wrote it to a real person. Watch one person open the page while you stay quiet. Collect the questions that arrive after someone has read it, because every one of them is a thing the page failed to say.

None of this means never buy traffic. A small paid test is the fastest way to learn whether strangers — not friends, not your existing list — respond to the thing at all, and [the first $100 will teach you more than a month of theorising](/blog/what-the-first-100-of-ad-spend-can-tell-you). Buying traffic is right when you have a page that converts and you want more of what already works, and it is right as a deliberate diagnostic. It is wrong as a substitute for reading a page you have not honestly looked at in a year. The failure mode is not spending the money; it is spending the money to avoid a harder decision.

When you do change the page, measure it in a way that can tell you something. Compare rates, not counts, or you will congratulate yourself for a busy week. Compare the same source before and after, because a shift in where visitors come from will swamp anything you did to the copy. Give it long enough to cover a full weekly cycle, since Tuesday and Saturday are different audiences wearing the same label. And change one thing at a time: rewrite the page and raise the ad budget in the same week, and you have bought yourself a bigger number and no information at all.

We build tools that draft page copy and pull your numbers into one place, and we will happily write you three versions of a headline in about a minute. That is the easy half, and it is worth being honest about which half it is. Software can produce the sentences. It cannot decide what is true about your product, or which of your customers you are actually talking to, and a confident draft aimed at the wrong person is worse than a clumsy one aimed at the right person.

So take one rule of thumb out of this: before you raise a budget, spend an hour with the page that budget will point at. If the hour is not worth spending, the money almost certainly is not either.