Forty percent

Instameal reached four hundred users, and forty percent of them placed an order. A good number. What it showed, and what was hiding behind it.

In 2024 I spent five months building Instameal. Foodtech from scratch: interviews with people, a customer journey map, a first version on a Telegram bot and a website, the first paying users. A team of eight: logistics, development, marketing.

On paper the result looks decent. Around four hundred users. Forty percent of those who signed up got as far as a first order.

In July the project ended.

A number you want to show off

Most services lose people right at the start. Someone signs up, looks around, closes the tab and an hour later forgets they were ever there.

When nearly every second person gets to a first order, it means the entrance is built right. The person understood what was on offer, didn't get lost in the interface and didn't give up at checkout.

That's a number you want to show. It goes on the first screen, gets said in meetings with a well-timed pause, goes into the deck in a big font.

I was proud of that number. And, as it turned out, proud of the wrong one.

What it doesn't tell you

Activation measures the first step. About everything else it stays politely silent.

It doesn't tell you whether the person will order a second time. It doesn't tell you what that first order cost you. It doesn't tell you how much money went into getting the person to the "sign up" button at all.

A first order is a promise. A second order is the check on whether you kept it.

You can be brilliant at first dates and still never see anyone for longer than a week. By first-date statistics you're doing wonderfully.

In food, scale is paid to the courier

In an ordinary app an extra user costs almost nothing. The server doesn't care whether a hundred people came in or a hundred and one.

Food doesn't work like that. Behind every order there's a kitchen, packaging and a person who delivers it. Every new order is a new cost, a real one, out of the wallet, not a line in the cloud.

So good activation in foodtech can sometimes mean something very unpleasant: you're just losing money faster. The better the funnel works, the more cheerfully you drive into the wall. With a lovely interface, happy users and an empty account.

What it looks like from the inside

Every team has a question nobody asks in the meeting.

The number is good, the chart is going up, everyone nods. And somewhere in the room hangs the question "and how many of them came back". Nobody asks it, and not because nobody thought of it. Nobody asks it because after that the meeting stops being pleasant.

A good metric works like a painkiller. While it's on the screen, nothing hurts. It hurts later, all at once and without warning.

A dashboard that praises you owes you fuck all. A useful dashboard spoils your mood a little. If looking at it makes you want to open champagne, you're most likely looking in the wrong place.

What I'd measure now

Three numbers, and all three are more unpleasant than forty percent.

How many people placed a second order within thirty days. Not the first. The second.

What one order cost the company, with the whole kitchen and the delivery, not what it brought in as revenue.

How many people came back on their own, without a discount, a push notification or the words "we miss you".

In my next project, Yonma Yon, I started from the other end: I measured day-thirty retention by cohort from day one. Not because I read a clever book. Because I had already once admired a pretty number while a project quietly came to an end.