The signup is not the finish line

Two channels brought you sixty signups last month. One cost $720 and the other cost $1,800, so you already know which one you are going to feed next month, and you decided it in about four seconds. Now here is the part your report did not show you: two weeks later, eight of the cheap channel's sixty were still using the product, and thirty-three of the expensive one's were. On the only number that pays your bills, the cheap channel cost you $90 a person and the expensive one cost $54. You are about to move money in exactly the wrong direction, confidently, using real data.

The reason this happens is not carelessness. The signup is simply where the measurement runs out. It is the last event your ad account, your analytics, and your landing page all agree on — after that the person disappears into your product, which is a different system with different numbers and usually a different person looking at them. So the finish line ends up drawn at the boundary of what is easy to see rather than at anything that matters, and once it is drawn there, everyone downstream starts optimizing toward it in good faith.

That is the actual damage, and it is worth being blunt about. Whatever number you report is the number you will buy more of. Ad platforms will find you more of the humans who do the thing you told them to count, and they are extremely good at it — give them "filled in the form" and they will go find the people who fill in forms, which is a real and distinct human trait with almost no relationship to the people who will still be here in March. Your own attention works the same way. A number on a weekly report becomes a target within a month, whether or not anyone agreed to make it one.

you will get more of whatever you decided to count.

So move the line one step further down and accept that the step is small. Not the whole way to revenue, which takes too long to steer by and drags in a dozen things marketing does not control. One step: the first thing a person does that, in your honest experience, tends to precede them sticking around. A few shapes it takes, depending on what you sell — imported their real data instead of the sample; invited a second person; came back on a different day without being emailed; placed an order that actually shipped; booked the call and showed up to it. Pick one. Not a weighted score of five things, because the moment it takes a formula to explain, nobody will believe it during an argument, and the whole point of this number is to settle arguments about money.

The honest tradeoff

The further down you move the line, the truer it gets and the slower it moves. Signups tell you something today and mean very little; the two-week number means a great deal and tells you nothing for two weeks. You do not resolve that tension, you staff both sides of it: keep the fast number as a smoke alarm — did the campaign run, is the form broken, did traffic fall off a cliff — and use the slow number for decisions, which are monthly anyway. This is the same discipline as picking one goal and one number, just applied one notch deeper, and it fails the same way if you let the list of numbers grow back.

You also do not need anything elaborate to start. If your product can report the moment it happens, hand that event to whatever you already use for reporting, so the source of the visit and the fate of the person end up on the same row. If it cannot, do it by hand: once a week, open two lists, write down which of the new people did the thing, and note where each came from. Four weeks of that is about forty rows in a spreadsheet, and forty honest rows will change a budget decision faster than a perfect setup you spend a quarter building and never quite finish. This is also the cheapest fix for the first-hundred-dollars-of-ad-spend problem, where the sample is too small to prove anything about clicks but plenty big enough to notice that nine of the ten people never came back.

What you find is usually specific and slightly embarrassing, which is a good sign that it is real. A channel you were proud of turns out to deliver people who sign up and evaporate. An ad that promises "free" recruits exactly the audience the word attracts. A discount code brings a wave of orders and a matching wave of refunds a month later, so the campaign was profitable in the report and a loss in the bank. None of this shows up while the finish line sits at the form, because at the form every one of those stories looks identical to a good one.

We build tools that pull your numbers from the places they live and draft the campaigns and emails that go out, and we can take that activation moment from your product and carry it back next to the source, so the two facts sit on one row instead of in two departments. What software cannot do is pick the moment for you. Choosing which single action means "this one is going to stay" is a judgement about your own product that comes from having watched a few hundred people either stay or leave, and any tool that offers to decide it on your behalf is guessing with more confidence than it has earned.

Here is the whole thing in one line, and it is worth putting on the wall next to your dashboard: a signup is a person who agreed to try, and you cannot run a marketing budget on agreements.