Decision-1 score question: order a Sume render queue by urgency

Yes, a Decision-1 score question is a reasonable way to order a render queue, as long as you use it for relative order only. Microsoft's page says the returned score is the probability-weighted average of the level indexes and advises preferring scores for relative ordering and thresholds, not as absolute, calibrated ratings.

How the score is built

You list levels from lowest to highest in criteria. With four levels the result runs from 0 through 3 and can land between levels. A brief the model finds borderline between Minor and Major might score 1.5, which is exactly what you want when sorting.

Order, then submit in waves

Sort your briefs by score and submit the top of the list first. Because Sume does not document a priority input, the order is your own: send what matters most while capacity remains, and hold the rest. The generation admission page explains how queued and processing jobs count, and jobs_wait or polling tells you when a slot opens.

Wording shifts scores, so keep the same question text for the whole batch and re-test it if you edit it.

Edge cases

Ties are common when scores are rounded. Break them with something you control, such as submit time. If a request's score is missing, because the call failed, put it in review and not at the head of the queue.

Re-score when the brief changes, because a changed state is a new question to the model.

• Break ties by time.
• Never default a failure to top priority.
• Re-score on edit.