A shop asked whether rain was good for business. The answer changed three times.

SECTION 01The question

A home and garden shop with a street frontage wants to know something simple. Do we do better when it rains?

It is a good question and it comes from somewhere real. The owner has stood behind that counter through a lot of wet afternoons and formed an impression. They want the impression checked, and they have a hunch about what to do if it holds: the umbrellas are already by the door, so put them on offer when the sky opens and pull people in out of the wet.

What follows is how that work actually goes. Three years of daily sales on one side. Weather on the other, which is free and takes about a minute to fetch.[1]

The answer changes three times before it settles, and each change is worth more than the answer.

SECTION 02The first answer: plus 31 percent

Flag every trading day as wet or dry. Average the revenue in each bucket. Compare.

Wet days come out 31.4 percent ahead.

The owner's impression is confirmed, spectacularly, and this is the moment where a lot of analysis stops and a chart gets emailed. It is also where the number is at its least true.

SECTION 03The second answer: most of it was summer

Rain is not scattered evenly through the year, and neither is a garden shop's trade. The wet months and the busy months are the same months.

So a plain wet-against-dry comparison hands rain the credit for everything the calendar was already doing. Every February Tuesday sits in the dry bucket dragging it down, and rain gets paid for the difference.

The first number out of any comparison like this is usually wrong, and it is wrong in a direction you can predict before you compute it.

The repair is not clever. Compare each wet day against dry days from the same few weeks, rather than against the whole year. Same season, same trading conditions, same part of the calendar.

Rain is now worth 8.9 percent.

Two thirds of the original finding was the seasons. That is the normal proportion, which is why this step is not optional and why nobody enjoys it.

SECTION 04The third answer: rain at 3am is not rain at 2pm

A daily wet flag is a crude instrument. It counts a storm that passed through at four in the morning, while the shop was shut and everyone was asleep, exactly the same as a downpour that started at two in the afternoon with the doors open.

Those are not the same event and there is no reason to expect them to do the same thing to revenue.

Hourly observations fix it, and they are as free as the daily ones. The relevant question is whether it rained during trading hours, which is a different column entirely.[1]

The effect is real and it is more than twice what the seasonal figure suggested, but it only exists on days when it rained while the shop was open. Rain overnight does nothing, or very slightly less than nothing.

That is a better answer and it is still not the useful one, because it says what happens without saying why, and every decision worth making depends on the why.

SECTION 05The cut that changed everything

Revenue is two numbers wearing a coat. How many people bought something, and how much each of them spent. A 19.8 percent lift can be either, or both, and they point at completely different businesses.

Fewer people come in when it rains. The ones who do spend far more.

That single line rewrites the whole question. The owner's plan, the umbrellas on offer to pull people in out of the wet, was built on the idea that rain brings people through the door.

Rain does the opposite. It keeps people away. The revenue lift is happening in spite of the footfall, not because of it, and it is happening because the people who do turn up in the rain are a different kind of visit: deliberate, unhurried, already committed to something, and in no rush to walk back out into it.

SECTION 06The umbrella trap

Now look at what the obvious plan would have cost.

The shop sells around 1,240 umbrellas a year at about $28. Seventy-one percent of those units, roughly 880 of them, already sell on wet days. At full price. To somebody who is already wet and already standing inside the shop.

An umbrella bought in the rain is the highest-intent purchase in the building. Nobody is shopping around. Nobody is waiting for a better offer next week.

This is the most common shape of expensive mistake we see, and it is never carelessness. It is a sound instinct, applied to a mechanism that turned out to run the other way. The instinct was worth having. It just needed one more cut of the data before it turned into a price change.

SECTION 07What goes in the envelope

Not a recommendation. Three doors, what each is worth, and what each costs to find out.

1. Leave the umbrella price alone and put a second item beside it. The wet-day customer is not price sensitive, they are time rich. They are already spending 36 percent more and they are in no hurry to leave. The cheapest test in the building, costs nothing, and the current margin stays intact while it runs.
2. Treat the missing 11.8 percent as its own problem. Fewer people is a door problem, not a price problem, and the levers are different: what the frontage looks like from across a wet street, whether there is anywhere to put a dripping umbrella down, whether the shop is findable to somebody deciding in ten seconds under an awning.
3. Leave wet days alone entirely. They are already the best days. This is the door the arithmetic in the next section argues for, and it is the one nobody asks about.

The owner knows which of these fits their shop and their street. We do not. What we can do is make sure the choice is made against the real mechanism rather than the plausible one.

SECTION 08The question nobody asked

Wet trading days are about 32 percent of the year. Running at 19.8 percent above the rest, they carry 36.1 percent of annual revenue.

Dry days carry the other 63.9 percent.

One point of improvement on a dry day is worth 1.77 times one on a wet day.

The shop spent its curiosity on the smaller and healthier half of its own business, which is completely reasonable, because the rainy day is the one that feels different when you are standing in it. Nobody forms a hunch about an ordinary Tuesday. Ordinary Tuesdays are 68 percent of the year.

That is most of what this work is. Not finding a number nobody could find. Finding out which question was worth asking, which is usually not the one that arrived.