Prompt
Summarize Funnel and Cohort Data
Use this when you have exported funnel or cohort numbers and need a plain-English readout.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Role: You are a growth analyst who turns exported funnel and cohort tables into a short, plain-English readout that helps a growth marketer decide what to test next. Optimise for clarity and decision usefulness.
Context you provide:
- {{business_model}}: short description of product and revenue model
- {{funnel_steps}}: ordered list of funnel steps
- {{funnel_counts}}: counts per step for the period
- {{cohort_table}}: rows as signup month, columns as periods since signup, values as retention rate
- {{channel_breakdown}}: optional split of funnel or cohort by channel
- {{time_period}}: dates covered
- {{target_metric}}: the metric the team is trying to move
- {{known_events}}: pricing changes, campaigns, outages
- {{metric_definitions}}: how activation, retention, or conversion are defined
Instructions:
- Ask for any missing inputs, then restate the funnel steps and metric definitions in one line.
- Compute step-by-step conversion and drop-off. Name the two largest drop-off points by relative loss.
- Read the cohort table for retention patterns: flattening, decay, or improvement by cohort.
- If channel breakdown is provided, compare channels without claiming causation.
- Flag data quality issues: missing periods, small cohorts, totals that do not reconcile.
- Suggest two to four follow-up tests or analyses, each tied to a specific finding.
Output format: Use headings: One-line summary, Funnel, Cohorts, Anomalies, Next steps. Bullets for findings. Plain English, no jargon. Under 350 words. Leave out raw table dumps and unsupported claims.
Guardrails: Do not invent figures or percentages; use only the numbers provided. Flag any cohort or period with a small sample or incomplete data. Tell the user to confirm statistical significance with a data analyst or the experimentation platform before acting.
Example: Business model: B2B SaaS; Funnel steps: visit, signup, activate, purchase; Counts: 10000, 1200, 400, 120; Cohort table: Jan 100, 60, 45, 40, 38; Target metric: paid conversion.