Aglet

Prioritize a partial rollout with divergent outcomes

A localized regression may be smaller than a global outage yet more urgent than older work because exposure is expanding. Make the queue decision around the affected cohort and the next rollout boundary. Record whether the priority is driven by harm, growth, or uncertainty.

Decide where the work belongs

  1. Measure exposed harm

    Estimate affected requests or users relative to the exposed cohort, and separate failed attempts from retries or duplicate reports. Identify whether the next rollout step would enlarge the population before treating expansion as safe.

  2. Account for reversibility

    Consider whether exposure can be held, narrowed, or compared with a stable control under existing release authority. Do not prescribe an unreviewed rollback. Record which safe decision points are available to the responsible owner.

  3. Set a rollout checkpoint

    Choose the next checkpoint at which evidence will be reviewed: before expansion, after a load threshold, or after a regional cycle. Name the result that permits progression and the result that requires investigation.

What to carry forward

Prioritization should leave the rollout at a named review point with a scoped owner and evidence rule. If cohort harm is low but comparison quality is poor, prioritize measurement rather than allowing aggregate health to stand in for confidence.

Keep the decision with the work.

Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.

Create an account See the product workflow