Diagnose a sudden drop in a key product metric
A structured investigation plan that rules out tracking errors and narrows down the real cause fast.
The prompt
Help me investigate a sudden change in a product metric.
Metric and definition: [e.g. daily signups, checkout conversion]
What changed: [FROM X TO Y, START DATE, GRADUAL OR SUDDEN]
Recent events: [RELEASES, EXPERIMENTS, PRICING CHANGES, CAMPAIGNS, OUTAGES, HOLIDAYS]
Breakdowns I can pull: [platform, country, channel, plan, app version, browser]
Data I already looked at: [WHAT I CHECKED AND FOUND]
Work through this in order:
1. Is it real? Checks for tracking breaks, pipeline delays, definition changes, bot traffic, and dashboard filters.
2. Is it external? Seasonality, day-of-week effects, marketing mix changes, competitor moves, platform or app store changes.
3. Is it internal? Map the timing against releases, experiments, and config changes.
4. Where is it concentrated? Which segments to slice first and what each pattern would point to (e.g. one platform only suggests a release bug; all segments equally suggests tracking or traffic source).
5. Is it a mix shift? Explain how to check whether the rate changed or just the population.
Return:
- A ranked list of hypotheses, each with the specific query or chart that confirms or rules it out.
- A short status update I can send stakeholders now: what we know, what we are checking, next update time.
Do not conclude a cause from my description alone; label each as a hypothesis until data confirms it.
Highlighted [brackets] are blanks to replace with your own details.
Fill in the blanks and use this promptMore Data & Analysis prompts
What to fill in
- e.g. daily signups, checkout conversion
- FROM X TO Y, START DATE, GRADUAL OR SUDDEN
- platform, country, channel, plan, app version, browser
- WHAT I CHECKED AND FOUND
How this prompt is built
This prompt follows the Few-Shot structure (Few-shot prompting). A few-shot prompt includes one or more worked examples of the input and the output you want, so the model can copy the pattern instead of guessing it.
Related prompts
- Explain my Excel model to a non-technical executiveTurns a complex spreadsheet or model into a plain-English walkthrough an executive can follow, trust, and question.
- Run a fair lending review of loan pricing and approval dataA structured fair lending analysis plan and readout that tests for disparities in approvals, pricing, and exceptions.
- Analyze my fleet's CSA safety scores and plan how to lower themExplains which safety categories are driving high scores, which inspections and crashes matter most, and what to fix first.
- Find what drove my store's weekly sales resultBreaks a store's sales variance into traffic, conversion, ATV, and UPT, then gives specific actions for next week.