A process for generating and prioritizing ideas

In this episode of Datadrivet, Joni Lindgren and Jasmin Yaya walk through the process Jasmin’s team uses to come up with ideas for new tests and then decide which ones to run. The whole point is to make the decision repeatable instead of leaving it to whoever speaks loudest in the room.

It starts open. Everyone on the team is invited to bring every idea they have, and every idea is welcome at this stage. Nothing gets filtered out on the way in. Once the ideas are on the table, the team clusters them into themes, or tracks, and votes on which tracks promise the most learning and the most value for the organization. That vote is how a long, unruly list becomes a short set of bets worth pursuing.

From there, each idea is written as a hypothesis in a fixed shape: by doing x, we will reach goal y. Forcing an idea into that sentence is a quiet test in itself. If you cannot name the action and the goal, the idea is not ready.

To rank the hypotheses, the team uses a modified ICE score with four parameters. Three are scored from 0 to 5: value, ease of doing it, and learning. The fourth, high risk, is scored as -1 or 0, so a risky idea is penalized rather than rewarded. The parameters combine into a single score from 0 to 100 percent, and that number sorts the list. The hosts reference the standard ICE model as the starting point and a shared Google Sheets template the team works from, with the scoring approach drawn from Daniel Hansson at Signific.

The takeaway is that good prioritization is a small, honest system. Gather widely, group, vote, write the hypothesis, score it the same way every time. The score does not make the decision for you, but it keeps the team arguing about the right things.

Listen to the full episode of Datadrivet for the full walkthrough of the template and the scoring.