Every deck on this site opens with one "How might we" statement. I treat it as the first checkpoint. Until the team, the users, the mechanism, and the outcome fit in one sentence, we have no shared problem to design against.
How might we do X for these users in a way that uses this mechanism, so that this outcome happens?
Each slot does a job. "How might we" leaves room for more than one solution. "For" makes you name a real audience instead of "users." "In a way that" names the lever you plan to pull, so the statement can’t hide behind a feature name. "So that" ties the work to a business or human outcome you can measure, and that outcome keeps the rest of the process honest.
American Airlines: Flight Maintenance System
"How might we create a flight maintenance scheduling system for pilots and engineering teams, in a way that reduces the friction of communication when placing requests for repairs, so that the efficiency of workload and labor is optimized?"
Kraft Heinz: Predictive Maintenance
"How might we deliver a suite of predictive maintenance capabilities and inventory management for factories and distributors, in a way that optimizes downtimes in manufacturing and increases efficiency in labor, so that Kraft Heinz can drive revenue growth?"
I move a framing statement forward once it passes three checks:
- Read it out loud. If it names a feature instead of a mechanism, it’s premature.
- Ask "so what?" about the outcome. If nobody answers with confidence, that slot is still a guess.
- Show it to someone outside the team. If they can’t repeat the audience and mechanism back to you, sharpen it.