Capture, cluster, choose
Good ideas die in three ways: never captured, captured but lost in noise, or beaten by whoever argued loudest. The Capture-Cluster-Choose pipeline fixes all three. Capture widely and cheaply, so ideas from the edge of the organisation reach the table. Cluster to find the themes and the duplicates, so you evaluate problems rather than personalities. Choose with explicit criteria: strategic fit, expected value, confidence, cost of pursuit.
Divergence before convergence
- Separate idea generation from idea evaluation; mixing them kills both.
- Score against the strategic bets, so "interesting" does not beat "important".
- Fund small experiments before big commitments, and let evidence promote ideas.
- Tell people what happened to their idea; silence is how the pipeline dries up.
Choosing ideas this way feeds directly into quarterly OKR setting: the winners become objectives and experiments, everything else waits with a reason attached.
Generation and adjudication are different jobs
Idea processes fail at two separate points: rooms that generate only variations of what they already do, and rooms that choose by advocacy strength, where the most senior or most enthusiastic voice wins. The two failures need different fixes, and conflating them is why brainstorming has a bad name.
The methodology connection
For generation, the three engines in possibility portraits: analogy, importing the logic of a structurally similar solution from any industry; trade-off, refusing a constraint everyone accepts; and anomaly, building for the customer who behaves "wrongly". Each produces options the current frame cannot. For adjudication, What Would Have to Be True: map each option's conditions, find the pivotal doubtful ones, test those, and let evidence rather than seniority decide. The chosen idea then becomes a bet with its conditions carried into the quarter's Key Result narratives.