PROMPT This AI Challenge
"How to Turn Unstructured Data Into Your Competitive Edge" featuring Warren Kucker
Every episode of the PROMPT This podcast includes an AI Challenge for the audience. Follow the instructions below to complete this episode's challenge.
Product roadmaps often get hijacked by the loudest voices, not the strongest signals. AI makes it possible to systematically audit customer feedback and tie feature requests to real business outcomes.
The challenge
Use AI to analyze customer conversations across sales, support, and customer success, then turn that raw input into a ranked feature backlog based on evidence instead of opinion.
What you feed the AI
Pull real customer data from a defined window, ideally the last 30 to 90 days:
-
Sales calls from discovery, demos, and negotiations. Simply export 20 or more call recording transcripts into a folder and then upload them into ChatGPT or your favorite Large Language Model (LLM).
-
Support tickets and escalations. Export these from your CRM using a report. Be sure to capture all of the notes, ticket classifications, related products and other important details used for decision making.
-
Customer success and renewal call notes. Whether this data is in your CRM, in a separate customer success system or in a spreadsheet, again collect all of the details into a single file for easy upload into the LLM.
Limit scope to a specific segment or motion, such as enterprise renewals or mid-market expansion, so results stay actionable.
What the AI analyzes
Ask AI to extract and standardize feature requests, merge duplicates, and separate true feature needs from bugs or edge cases. Then have it score each request across three dimensions:
-
Frequency: how often it appears across conversations
-
Revenue impact: deals, expansions, or ARR tied to the request
-
Churn risk: renewal pressure, downgrades, or at-risk accounts
Weight revenue and churn more heavily than raw frequency to avoid building popular but low-impact features.
Output
A ranked, defensible backlog where every item includes example quotes, affected accounts, and a clear business rationale. Anything without evidence drops to the bottom or into a “needs more data” bucket.
AI Prompts
Final Thought
The squeaky wheel is not your product strategy. Run the audit, force receipts, and let frequency plus dollars plus churn decide what matters. When product uses the same evidence language as sales and success, roadmap conversations get shorter and outcomes get better.
