The vote and the evidence picked different ideas
Thought leadership by successful women topped the decision matrix (cost, time, impact, scale, feasibility), but came third in the weighted vote. The ideas people liked most scored worst on cost, time and feasibility.
Taking over a project mid-stream
We worked as an agency taking over a project mid-stream. The client's research had already been done by another team, "So… what's the problem?", on why girls and women drop out of sport, and it was handed to us to turn into one idea worth prototyping.
How we generated ideas, and how we chose
Generate
- Problem statement, report insights and "how might we" questions.
- Brainwriting across four panels. My seed was panel 2: design sports outfits, a marketing campaign or an intervention that improves girls' confidence in sport.
- A reverse brainstorm (the worst ideas, flipped) and role storming as three characters.
Choose
- Dot voting, then super voting with weighted votes for "managers".
- An idea portfolio.
- A decision matrix scoring cost, time to implement, impact, scalability and feasibility.
| Idea | Weighted votes | Decision matrix |
|---|---|---|
| Co-designed sports outfits | 7 | 12 |
| Marketing campaign for confidence | 7–8 | 18 |
| Thought leadership by successful women | 3–4 | 21 |
The matrix picked thought leadership: talks and mentoring that show girls a route through, and cheap to start.
Not ready to build
In the design documentation we told the client the ideas weren't good enough to take into implementation: the handed-over research needed more investigation before ideating again. That's an honest result, and it traces back to the gap I'd flagged in the handover.
The difference was already in our responses
In the brainwriting sessions (panel 3, same six clusters), the others answered with solutions and I answered with questions about what we didn't know. These are the real responses from the session records:
| Cluster | A solution response (from the team) | My response: a discovery question |
|---|---|---|
| Sharing sessions | "We also can create an online web or app to do this" | "Understand more about what kind of interventions females believe would solve their confidence issue" |
| Thought leadership | "Discussion/seminars about thoughts" | "Understand what the leaders have successfully done to improve their own confidence" |
| Beginner sessions | "Make it extra fun" | "Understand whether it is the level of sports capability that is preventing them from engaging in sports more" |
| Parental education | "Events that include parents" | "How parents might play a role in helping improve the confidence of their daughters, nieces, granddaughters" |
| Mentorship | "Find funding for good mentors" | "Other ways of ensuring women feel guided with overcoming confidence issues related to capability" |
The detective in our role storming said the same thing: "Follow the dropout curve" and "The university dropout reveals the truth". The discovery questions were on the table, but voting and the decision matrix only had a way to rank solutions, so they dropped out.
Ideate the discovery, not the solution
With the handover in doubt, the better use of the ideation sprint was to design how to fix the research.
- Ideate a discovery strategy
Use the same tools on questions like "How might we reach the girls who stopped?" and "Which assumptions in the handover need testing first?"
- Redo the missing work
Run the research with the people the brief is about, the ones who dropped out.
- Route it at a gate
If the new findings back the handover, pass it forward to implementation with the ideas re-ranked. If they don't, pass it back to discovery with a clear brief of what's missing.
The two approaches side by side
Same tools, pointed at a different question.
| What we did: ideate the solution | The redo: ideate the discovery | |
|---|---|---|
| Starting question | How might we improve girls' confidence in sport? | How might we find out why girls who stopped, stopped? |
| Assumes | The handed-over research is right | The handed-over research may have a gap |
| Brainwriting produces | Interventions: outfits, campaigns, mentoring | Research routes: who to reach, where, by which method |
| Reverse brainstorm asks | How could we make girls quit sport? | How could we make sure we never hear from the girls who quit? |
| Role storming as a detective | "Redesign the infrastructure, not the girls" | "Follow the dropout curve: interview at the age it peaks" |
| Voting and the matrix rank | Ideas, by cost, impact and feasibility | Research routes, by reach, cost and how much they'd change the answer |
| What gets prototyped | An external prototype: the product (a mock-up outfit or campaign), for the client's users | An internal prototype: the research (a pilot of the discovery method itself, UXR), for the agency and the client |
| What the test tells you | Whether people like the product | Whether the method reaches the right people and gets honest answers |
| Where it loops back | Into discovery, through product testing | Into discovery, with a method proven before it's scaled |
| The risk | Testing a product before checking the problem | No product yet, only a better-founded brief |
Both routes end in a prototype, and both loop back into discovery. An external prototype is entrepreneurial: it tests the product on the people it's for. An internal prototype is intrapreneurial: it improves our own way of working before we rely on it.
That way the agency hands the client a decision it can act on either way, not ideas built on a gap.
A vote measures enthusiasm, not evidence
I now rank options by what they cost, what they'd change and how sure we can be, which is the confidence rating I used later in case study 05.
Then → Now
Then: Ideating on a timer from thin material didn't work for me, and I don't think I'm alone. Now: I share material before group sessions so everyone gets solo thinking time first.
Then → Now
Then: I didn't trust dot voting. Now: I score options against criteria and say how confident we are.
Then → Now
Then: The design process felt like a straight line. Now: I treat it as a loop: any stage can need its own small ideate, test and deliver.