Understand what a count represents.
A theme with many records deserves attention, but the count may combine repeated messages, duplicated imports, or several notes from the same conversation. Before comparing themes, inspect what is being counted.
Even a clean count reflects the channels and audience available to you. A quiet group may be underrepresented, and a small number of sources may describe a severe blocker. Frequency is one input to a decision, not the decision itself.
Compare the problem in context.
Discuss the task, the audience, the severity of the difficulty, and the current product strategy. Then separate the evidence from assumptions about effort or commercial impact. This lets the team challenge each part of the argument independently.
A simple review table can be more useful than a single opaque score. It should preserve uncertainty rather than converting every unknown into an apparently precise number.
| Dimension | Question |
|---|---|
| Evidence coverage | Which distinct sources and audiences are represented? |
| Problem context | What task is blocked or made harder? |
| Strategic fit | Why is this relevant to the current direction? |
| Effort assumption | What needs technical investigation? |
Record why the team chose the next step.
A priority conversation should leave a short record. State the chosen action, its owner, the evidence that mattered, and the uncertainty that could change the decision.
The next step may be discovery rather than implementation. A bounded research question is often more defensible than approving a solution when the team has not agreed on the problem.
Use counts to start a question. Use evidence and judgment to decide what follows.
Make the decision revisitable.
Attach a review condition: a new source group, a technical finding, or a change in strategy. Reconsidering a decision in light of better information is easier when the original reasoning remains available.
This makes prioritization a maintained practice rather than a list that gradually loses its context.