I design tools for people who work with data, and I write about how they get made.

On a boat outside the Koster Islands, a marine biologist showed me the tide table she had folded into her notebook. Several cells were blank. I asked whether the data was missing. “No,” she said. “There’s nothing to measure there yet.” She knew exactly what the blank meant. Our app would have called it an error.

We were testing Tidepool, a field app for recording coastal observations. I had designed a perfectly tidy empty state: a small illustration, a friendly heading, and a button to add an observation. On land, it made sense. On the water, it collapsed several different situations into one cheerful screen.

First, name the kind of nothing

A site might have no observations because nobody had visited it. It might have observations that were still on another device. Or a researcher might have visited, looked carefully, and found no eelgrass. That last case was data. Treating it as an invitation to “get started” erased the work someone had already done.

“Nothing found” and “nothing recorded” tell two very different stories.

We stopped drawing screens and started writing down the states. The list was short, but it changed the conversation with the developers:

observation.status:
  not_visited  # no field visit yet
  not_found    # visited; species absent
  pending_sync # recorded on another device

Each state needed its own sentence and its own next step. “No visit recorded. Add a field visit.” “No eelgrass found on 6 October.” “Waiting for observations. Connect when you’re back on shore.” We could remove the illustration because the explanation was finally doing its job.

Design for the person who arrives next

The useful test wasn’t whether the original researcher understood the screen. It was whether a colleague opening that site three weeks later could tell what had happened. Had somebody looked? Was the device waiting? Should they go out again? The interface needed to preserve that context without asking anyone to remember it.

I now review empty states alongside populated ones, with real examples of how a record becomes empty. I ask what we know, what we don’t, and what the person can do next. Sometimes the right response is a button. Sometimes it’s a timestamp. Sometimes it’s a plain sentence that makes absence legible.

The tide table stayed on the boat. The distinction came back with me.

Written in Gothenburg, with a notebook still smelling faintly of the sea.