One bad Tuesday is not a verdict

A small online seller in Doha opens her sales on a Wednesday morning and sees that Tuesday was almost empty.

By lunchtime she has decided several things. The new product is a flop. The advert has stopped working. Customers have found somewhere cheaper. She should cut the price, change the photos and probably stop the advert altogether.

That is a long way to travel on one day of data.

It might be any of those things. It might also be a public holiday in a neighbouring market, a payment page that was slow for six hours, a payday that fell on Thursday this month, or simply a quiet Tuesday, the kind every shop has had since shops began. One figure can tell you something moved. It almost never tells you why, and it certainly does not tell you what to do.

The trouble with leaping is that you land somewhere confident and then defend the landing. You stop looking, because you already know.

Walking is slower and much more useful. Check whether the drop is real or a glitch. Look at the next day. Look at the same day last month. Ask two customers. Change one thing, not five, and watch. Each step is small, and each one rules something in or out. By the time you arrive, you understand the ground you crossed, and the answer is more likely to be right.

There is one place where jumping is fine. When a customer is rude, or an order is cancelled for no reason, or a review stings, leap straight to the kindest explanation. She was having a bad week. He misread the size chart.

Those leaps cost nothing.

Save the long walk for the decisions that do.