Three months later, there's another chart to build.
This one shows anxiety check-ins over time. The team wants people to look back at their responses and understand what's changed.
While exploring the chart component, the designer finds a link to the decision record we've been writing about. It explains why the mood chart includes a short summary and access to individual entries. It also explains where that approach applies and what the original team didn't test.
They have a useful starting point, but they still need to think about their feature. Anxiety check-ins aren't mood ratings and missing responses could make a summary misleading. The earlier work doesn't settle this, but it helps them narrow the question. How should the summary describe a period when several days have no entries?
So they explore that, test the chart with screen reader users and revise the wording. Before shipping, they add what they learned about missing entries to the decision record, including when that guidance applies.
They still design the feature, develop and test it properly. They don't copy over a finished answer and call it done. They can't. But they also don't spend days rediscovering why a summary alone isn't enough or how someone could look at their responses without actually seeing the chart.
And that's the difference I wanted from the beginning.
The decision record gives people a place to start and their work then makes it more useful for whoever comes next. And whoever comes next still has decisions to make. They just have better information to make them with.