How to write "How might we" questions that lead somewhere
Run this method in IdeaHone
Creates a space with the entities and scoring criteria of this method, ready for your next session.
"How might we increase customer loyalty?" gets a wall of generic ideas. "How might we get first-time buyers to come back within 30 days?" gets ideas you can test next week. The How might we question is the most used and worst-written sentence in innovation work. This guide shows you how to write one that leads somewhere.
Why the question decides the workshop
Everything in an ideation session hangs on one sentence. A vague one gets filled with whatever is already in people's heads: the app, the loyalty programme, the newsletter. A precise one forces the group to think about one person in one situation, and the ideas get specific on their own.
The three parts of a usable question
- A specific person or situation. Not "customers" but "first-time buyers", not "employees" but "new field technicians in their first month".
- A desired change. What should be different afterwards: come back, finish the form, stop calling support.
- A measurable or observable outcome. Something you could notice: within 30 days, without a second visit, on the first attempt.
Miss one part and the room supplies it, each person differently.
The Goldilocks test
Too broad. If it could be any company's question. "How might we improve the customer experience?" fits a bank, a bakery and an airline.
Too narrow. If it already contains the solution. "How might we build an app that reminds customers to reorder?" is a brief with a question mark. The group can only decorate the app.
Right: ten people could give ten different kinds of answer, all about the same person in the same situation.
Five bad questions, rewritten
| Before | After | What changed |
|---|---|---|
| How might we increase customer loyalty? | How might we get first-time buyers to come back within 30 days? | Person and outcome replace a KPI. |
| How might we improve onboarding? | How might we help new field technicians finish their first job alone in week one? | One situation with a finish line. |
| How might we build a chatbot for support? | How might we let customers solve a delivery problem without waiting for an agent? | Solution out, customer's situation in. |
| How might we increase engagement and reduce churn? | How might we get users who have not logged in for 14 days to return? | Two questions became one. |
| How might we use AI in our processes? | How might we cut the time a case worker spends re-typing forms? | Technology wish became a person's pain. |
Where the question comes from
A usable question comes from an observation, never from a solution someone already has: a sentence from a customer interview, a recurring support complaint, a number that surprised you. If the sponsor "wants an app", ask what they saw or heard that made them want it. That observation is the question; the app is one answer.
How to write it
- Start from one observed problem. Write the observation on the wall as it happened: "Customers who buy once rarely buy again." No interpretation yet.
- Name the person. Who is in that situation? As specific as your research allows: first-time buyers, not customers.
- Name the change. What should be different for that person, and how would you notice?
- Remove the solution. Cross out every noun that is a product, feature or channel. If the question collapses, you had a plan, not a problem.
- Check it is open enough. Ask yourself for ten different ideas in two minutes. Only variations on one idea: widen it. Nothing in common: narrow it.
- Write three variants and pick with the sponsor. One broader, one narrower, one from another angle. Choosing between three takes ten minutes; approving one is politics.
Write the final question on the wall, large enough to read from the back of the room, and read it out loud before every round. Groups drift after twenty minutes; the question brings them back.
Hidden flaws in a How might we question
- The question that hides a decision already made. "How might we move our customers to the new portal?" Only the method is open. Say so, and run a planning session instead.
- The "and" question. "Increase engagement and reduce churn" is two questions. Split them or pick one.
- The question nobody in the room cares about. If the group has never met that person, bring a recording, a quote or the person. Otherwise you get polite ideas.
- The company's KPI instead of the customer's situation. "Increase conversion by 5 %" is your problem, not theirs. Ask what the customer is trying to do when you lose them.
Once the question is on the wall, the 25 brainstorm questions keep the ideas coming, and the 90-minute ideation workshop agenda shows where it sits in the session.
Capture every idea while people talk
IdeaHone listens in your session and turns what people say into idea cards, questions and decisions, live on a shared board.
What is a How might we question?
A short, open sentence that frames a problem for ideation: a person or situation, a desired change and an observable outcome, with no solution in it.
How specific should a How might we question be?
Specific enough that it could only be your question, open enough for ten different kinds of answer. Fits any company: too broad. Names a product: too narrow.
How many How might we questions should a workshop have?
One per ideation round, usually one to three per workshop. More and the group never gets deep on any; pick the one the sponsor cares most about.
