30 coffees, 30 restaurant owners, one full notebook

Before coding anything, we set ourselves a limit: 30 interviews with restaurant owners before opening the editor. Taverns, grills, cafés, a couple of fine-dining kitchens, a three-location franchise. No rigid questionnaire; coffee, a notebook, and two opening questions: "What tool do you use today?" and "What drives you up the wall every week?".

We thought we would walk away with a long, scattered list. We walked away with three sentences that repeated almost word for word.

The three sentences we wrote in big letters

These came up in more than half of the conversations:

1. "TheFork charges me too much." It was not just the monthly fee; what hurt most was the per-cover commission. Restaurants with a healthy average ticket felt they were paying a disproportionate toll for a guest who often already knew them.
2. "I can't find a tool that's actually in Spanish." They emphasized "actually". Half-translated apps, English-only support, documentation full of phrases nobody would say in conversation.
3. "My server has to be able to use it too." Zero learning curve. If the kid coming in for an afternoon shift cannot figure it out in five minutes, they are sticking with the notebook and the landline.

Other things that came up more than once

Not as repeated, but mentioned often enough to take seriously:

• Reservation reminders by WhatsApp or email that do not feel like spam.
• Keeping the restaurant's menu up to date on the website without having to call the cousin who knows computers.
• Knowing who the guest is when they arrive: are they regulars, do they have allergies, did they celebrate a birthday last time.
• Anything, literally anything, against Friday-night no-shows.

What we decided not to do

There were requests we did not pick up. Some because they were too specific to a single venue; others because they crossed into POS, inventory, or payroll territory, and we wanted to stay in a different lane. More on that in the next post.

What we did do was paste those three sentences on the office wall. They have been there since June. Whenever we are unsure if a new feature is worth building, we hold it up against that wall. If it does not answer one of the three, we shelve it.

The interviewer bias

An honest note: in the first five interviews, we asked leading questions. Things like "don't you miss a metrics dashboard?". People said yes out of politeness. From the sixth onward, we changed the script: we let people talk and only asked "and why is that?". The answers got dramatically better.