The Unreliable Narrator

I asked the accountant what the hardest part of her month was. She thought for a moment and said it was the clients who sent documents late. She said it with the certainty of someone who had answered the question in her head before I asked it.

A week later I sat with her through a Monday morning. The late clients were a nuisance. They cost her ten minutes of chasing, spread across the morning, and the chasing was mostly a matter of forwarding a template she had saved. The late clients were not the hardest part.

The hardest part was the counting. She opened the spreadsheet, looked at the grid of green and red cells, and tried to remember which clients she had already emailed that week and which she had not. The spreadsheet did not track the emails. Her inbox did, but her inbox was sorted by date, not by client. So she held the list of sent emails in her head and compared it, cell by cell, against the grid. She did this for perhaps twenty minutes each morning, and by Wednesday the list in her head was long enough that she sometimes emailed a client twice and apologized.

She had not mentioned this when I asked her the question. She had not mentioned it because she did not think of it as a task. It was the shape of the morning, and the shape of the morning was not a thing you named. The late clients were a thing you named, because they had a villain.

This is the problem with asking.

I studied anthropology before I wrote code. The first lesson of fieldwork is that people are unreliable narrators of their own lives. Not because they lie. Because the act of describing a day compresses it, and the parts that compress out are the parts the person has stopped noticing. A farmer will tell you about the harvest and leave out the walk to the well. A weaver will tell you about the loom and leave out the hour spent sorting thread. The walk and the hour are the life. The harvest and the loom are the story.

A good ethnographer learns to distrust the story. She asks her questions, writes down the answers, and then sets the answers aside and watches. What she sees rarely matches what she was told. The gap between the two is where the real work of the village lives.

Product discovery, as it is usually practiced, asks the questions and stops there. A founder interviews ten users, writes down what they say, and builds the thing they asked for. The thing they asked for is the story. It is the harvest and the loom. The user receives it, uses it once, and sets it aside, because the thing she needed was the walk to the well, and no one asked her about the walk, and she would not have known how to describe it if they had.

The walk is invisible to the person taking it. This is the whole difficulty. A task that has become automatic no longer feels like a task. It feels like weather. You do not complain about weather to a stranger with a clipboard. You complain about the late clients.

To find the walk, you have to be there when it happens. You have to sit in the corner of the office, or the kitchen, or the shop, and watch a person do her work until she forgets you are watching. Then the walk appears. She stands up, crosses the room, checks a thing, comes back, and sits down, and she does not mention it because it did not happen to her. It happened around her, the way weather happens.

The interview is a weak instrument. It catches what the user has already framed as a problem, which is almost never the problem worth solving. Observation is a stronger instrument. It catches what the user has stopped framing at all.

I am not against asking. You have to ask to get into the room. You have to ask to learn the names of things. You have to ask when you have seen something and do not understand it. But the asking is a doorway, not a method. The method is what happens after the asking stops.

The cost of treating the interview as the method is a product built for the story. It will look correct on a slide. It will match the quotes in the research deck. It will fail quietly in use, because the user will open it, look for the walk, and not find it.

The accountant told me the hardest part was the late clients. I built her a tool that handled the late clients, and I also built her a tool that handled the counting, because I had seen the counting. She uses both. She thanks me for the first one. She loves the second one and does not quite know why.

She does not know why because she never saw the counting either. It was her weather.