Non‑Generic Ideation Within a Restricted Business Setting
Year: 2026 | Industry: Fintech | Region: Japan | Role: UX Strategist/Prompt engineer | Team: UX Research | Deliverable: Ideation GPT (Chat GPT)
- Getting stuck in the ideation phase is something every designer is familiar with. While LLMs can help generate ideas, factors like time pressure, business metrics, and stakeholder preferences often limit the feasibility of these ideas.
- The UX research team has developed ideation formats before, but designers found them too complicated and too academic. It was therefore essential to keep the threshold for GPT usage as low as possible, while ensuring that the results surpass regular ideas in breadth, effectiveness, and feasibility.
- The result is an ideation GPT that only requires one of the following inputs: a business goal, a user problem, a business problem, current user behavior, or a quick draft of an idea.
- The GPT maps input against factors such as behavioral shift, user motivation, frequency potential, revenue potential, and execution risk.
- After mapping, the GPT generates five ideas:
1. Quick win: A small, fast improvement.
2. Safe scaler: Takes something that already works and scales it up.
3. Behavioral changer: Shifts a user habit.
4. Repeat behavior: Encourages recurring usage.
5. Bold bet: A new, risky idea.
- Most designers on our team use the GPT as a starting point for ideation. They don’t copy and paste the results directly into a slide. Instead, they use snippets and build on them. From there, designers use other GPTs to identify a target group or assess business inpact.
Note: The five ideas follow plausible, common patterns based on similar products. The UX research team recognizes that these patterns are still assumptions rather than validated facts, and that they cannot predict real-world impact.
Special thanks to the UX research team: Beau Tana, Pallavi Varma, Jane Susiriwatananont.