notes · thoughts ·
In operations screens, state comes before buttons
Looking back at ordering and assignment screens, the hard part was deciding what each state permits and what follows failure. UI conditions should be explainable as business rules.
Describing admin work by feature names leaves tables, filters, modals, and Excel downloads. But looking back at ordering and assignment screens, what I kept checking during implementation was what someone could do in the current state.
The same button means different things depending on whether an item is pending, on hold, or already assigned. A successful API call after a click is not the end. It connects to how the list and detail views should change and where a user should be able to retry if it fails.
Even the reason it is disabled
Disabling a button prevents an action. But an operator needs to know why it is unavailable to choose what to do next. The next place to check differs depending on whether the cause is the current state, missing permission, or incomplete input.
Processing several items at once is more complicated. If only some succeed, users need to distinguish what succeeded from what needs another attempt. Reusing the success message for a single item may not be enough.
When building an operations screen, I want to write down the states and available actions first, including what changes on success and what remains on failure. Then I can think about how tables and modals present those rules.
If the same rule appears separately in conditions in the list, detail view, and confirmation modal, it is easy to fix one and miss another. I should decide what belongs together by whether it shares a business rule, rather than only by how components look.
A well-rendered table alone does not tell me whether an operations screen is complete. I also need to see whether operators can understand the current state, choose their next action, and continue after failure. Rereading my work history, I found many of my recurring fixes in those spaces.