Most groups draw the first figure from the runs that finished cleanly. The abandoned setups stay in a different notebook, a chat thread, or nowhere. When a later colleague asks why a temperature band was dropped, the chart has nothing to say.
A reconstructable run sheet does not need to look like a quality-system form. It needs a start condition, the intended observation, what actually happened, and why the run ended. Those four lines are enough for a visualization briefing months later.
Start condition means the settings a new person would have to copy: materials lot if it matters, instrument identity, and the planned duration. Intended observation is the question the run was meant to answer, written as a sentence, not a project code. What actually happened is the sequence, including the interruption. Why the run ended is the sentence groups skip — completed as planned, stopped for a fault, or set aside because the question changed.
Failed runs belong on the same family of sheets as successful ones. If they live in a separate pile, they will not appear in the next stakeholder pack, and the pipeline will look healthier than the bench felt.
Paper and a local tracking utility can share this structure. The sheet is the source; the utility is a second copy that can be sorted. Neither should invent fields the bench will not fill. If a field is empty on three successive runs, remove it rather than scolding the authors.