Five things a berry farm taught me about building software

I built a timesheet system for a berry farm in Halikko, Finland. Over 300 seasonal workers use it now, in five languages. It was the first time I built something for people who have no choice but to rely on it, and I got a few things wrong along the way. These are the lessons I would keep.

1. Talk to the people who use it first

The formula for calculating hours changed several times, and every change came from something a worker or someone in payroll told me. I could not get it right until I understood the paper forms they had used for years.

2. Keep the calculation in one place

Early on, the frontend had its own copy of the formula, and it slowly drifted away from the backend. That caused a real bug. Now the calculation lives only on the backend, so every phone shows the same result.

3. Build for where people really are

Workers open the app in fields with poor signal, on cheap Android phones, in languages other than English. Every decision had to work under those conditions, not on my laptop.

4. Start simple

It began as a basic timesheet calculator. Batch recording for supervisors, payroll verification and the berry-picking form were all added because someone asked for them.

5. Meet people in person

The project moved forward because I sat down with Lina, Aleksi and Johanna, not because of emails and messages.

Before Rannikon, workers filled in four paper forms by hand and payroll spent days checking them. Now that check takes minutes. The full write-up, with screenshots, is in the case study. Read the case study.