Better wins the week

Two claims go into the same creative test.

The first says cheaper, faster, bigger, best in the city. The second says who the thing is for and what it refuses to do. No delivery outside the neighbourhood. One size range. Made this morning or not sold at all.

By the third day the first claim is ahead. It is almost always ahead, and the ranking gets read as the market answering the question.

The reading is wrong, because of who a click test can reach. It recruits from the people already shopping the category this week, and to somebody already comparing, cheaper is the sentence they came looking for. The better claim is not the more persuasive one. It is the one matched to the only audience the test has access to.

The second claim is aimed at somebody who is not in the market at all. They read the refusal, decide that sounds like a place for people like them, and turn up four months later when the need arrives. No fortnight of data can see that person, because the window closes before they exist.

So the test has a favourite, and its favourite is always whichever claim converts demand that was already there.

Run it every two weeks for a year, kill the losers, scale the winners, and the account becomes a well-optimised argument about price with a full set of numbers behind it that are all true. None of them answers the question the business was asking, which was who else could be brought to want this.

The different claim never gets a second reading either, because it lost twice in week one, and losing twice is enough for a line to stop being written.

Better is measurable in a week. Different is measurable in a year. The account only ever budgets for the week, so it never finds out what the other claim would have done, and it files the absence as proof that different does not work here.