How I Used AI to Analyze Product User Interviews and Get a 30-Day Plan
I ditched manual product interview analysis & automated it! Otter.ai transcribed voice memos, and ChatGPT extracted core themes, cross-interview patterns, & a prioritized 30-day product improvement plan. See how I simplified the effort with AI.
Historically, I've conducted product user interviews using a rather manual process followed by all the work required to extract meaningful guidance. As part of this AI journey, it made sense to simplify the effort using AI tools. Here's what I learned.
The Interviews
There is a rigorous approach to proper user interviews - defining research goals, recruiting targeted participants, crafting open-ended questions, etc. - all followed by rigorous analysis to extract qualitative and quantitative data.
Given that Burrow is more of a hobby to learn AI, my interviews were definitely not that structured. I reached out to some neighbors and friends to walk through the app while I recorded their feedback using voice memos. I did apply some best practices to the sessions, but it was pretty casual.
As I reviewed the results and tried to extract meaning, I realized there might be a better approach.
The User Interview Analysis
To analyze the interviews, I:
- signed up for otter.ai - an AI transcription tool
- uploaded the voice memos
- had it transcribe the sessions
- assigned a few lines of conversation to the appropriate participant so it could learn everyone's voices and complete the attribution on its own
- exported the transcripts to import into ChatGPT
- had ChatGPT perform a cross-interview analysis to identify recurring themes and extract insights
Given the level of effort, the results were amazing. It returned:
- Core friction themes
- Strategic synthesis across all of the interviews - highlighting key insights
- Cross-interview pattern analysis
- A summary for MVP readiness, including a prioritized listing of things to address
- a 30-day plan for product improvements
Why ChatGPT Worked
ChatGPT did have some advantages that made it more powerful. One of the biggest items is that it is well aware of what Burrow is, and we've used OpenAI Codex to conduct code reviews.
Another thing to note is that in a prior round of user interviews, I uploaded voice memos directly to ChatGPT to try to streamline the process and potentially capture insights from voice inflection. That did NOT go well. After 24 hours of trying to process the information and telling me it could do the task, it gave up. Hence, the inclusion of otter.ai this time.
Conclusion
For quick analysis of user feedback, this was a great approach. In the future, I'll skip the voice memos and record product user interviews directly in something like otter.ai, saving me a step.
