Interview Questions › Product Strategy

Double weekly users of an AI image tool with 3 engineers

How would you grow weekly active users of an AI image creation tool (such as ChatGPT image generation) from 175M to 350M in 3 months, with a team of only 3 engineers?

Context

The tool went viral at launch and usage is now flattening. Compute is expensive. The team is tiny, so every bet must be cheap to build.

Clarifying questions to ask first

• Is the problem new users or people not coming back?
• Do we care about paid conversion or only usage?
• What compute budget do we have?

How to approach it

1. Clarify what counts as an active user and where growth stalls (acquisition, activation, retention)
2. Segment users by intent and pick where doubling is realistic
3. Choose a few cheap, high-leverage bets that fit 3 engineers
4. Define metrics and guardrails (compute cost, safety, copyright)

Framework: Goal & definition -> Funnel diagnosis -> Segments -> Cheap high-leverage bets -> Metrics, guardrails & risks

Sample answer

• Retention, not acquisition, is the bigger lever after a viral launch
• Casual users drop off because they do not know what to make next
• With 3 engineers, prefer prompts, templates and sharing loops over new models

1. Bets: shareable templates and remix loops, weekly creative prompts, re-engagement for lapsed users
2. Cut: anything needing new model training or complex backend
3. Metrics: WAU and week-4 retention; guardrails: compute cost per active user, policy violations, copyright complaints

Focus on retention and sharing loops (templates, remixing, weekly prompts) that 3 engineers can ship. Measure WAU and week-4 retention, guarded by compute cost and copyright or safety complaints.

What interviewers look for

• Funnel diagnosis
• Prioritisation under constraints
• Cost and safety guardrails
• Realism of the target