Cohort Analysis Guide
The Cohort Analysis Guide helps you build a complete retention analysis from raw user or transaction data. Instead of spending hours structuring cohort tables and interpreting patterns manually, you provide your data and business context, and receive a fully structured cohort analysis with visual tables, trend explanations, and concrete recommendations.
Product managers tracking user retention, growth marketers measuring campaign effectiveness, subscription businesses monitoring churn, and SaaS teams analyzing feature adoption use this prompt. It applies to any scenario where you need to group users by their start date (or another defining event) and track their behavior over subsequent time periods.
This prompt produces significantly better output than asking "analyze retention in my data" because it enforces a rigorous cohort methodology: explicit definition of the cohort criterion, consistent time period bucketing, normalized percentage tables alongside absolute numbers, and separation of observation from interpretation. It also requires the AI to call out the specific cohort that deviates most from the average, which surfaces the actionable insight rather than burying it in a generic summary.
This prompt is just the starting point
Score it with AI, optimize it with one click, track versions, and build your prompt library.
The Prompt
Perform a cohort analysis on the following data to identify retention patterns, behavioral trends, and actionable opportunities: **Business Context**: - Product/Service: [DESCRIBE YOUR PRODUCT, e.g., "B2B SaaS project management tool, $49/month subscription"] - Key Metric to Track: [METRIC, e.g., "monthly active usage" / "repeat purchase" / "subscription renewal"] - Time Granularity: [weekly / monthly / quarterly] **Cohort Definition**: - Group users by: [COHORT CRITERION, e.g., "month of first signup" / "acquisition channel" / "first purchase date" / "onboarding completion week"] - Observation window: [TIME RANGE, e.g., "January 2025 through December 2025, tracking each cohort for 6 months after signup"] **Data**: ``` [PASTE YOUR DATA HERE: user IDs with event dates, transaction logs, or pre-aggregated cohort counts. Include at minimum: user identifier, cohort-defining event date, and subsequent activity dates or flags.] ``` **Analysis Requirements**: ### Step 1: Cohort Table Construction - Build a cohort retention table with cohorts as rows and periods since the cohort-defining event as columns (Period 0, Period 1, Period 2, etc.). - Show two tables: one with absolute user counts and one with retention percentages (normalized to Period 0 = 100%). - Clearly label what each period represents (e.g., "Period 2 = the user's third month since signup"). ### Step 2: Pattern Identification - Calculate the average retention rate for each period across all cohorts. - Identify the **steepest drop-off period** (where the largest percentage of users are lost) and quantify it. - Flag any cohort that deviates more than [THRESHOLD, e.g., "10 percentage points"] from the average retention curve. Explain possible reasons. - Identify the **best-performing cohort** and the **worst-performing cohort**, with specific numbers. ### Step 3: Trend Analysis - Compare early cohorts versus recent cohorts. Is retention improving, declining, or stable over time? - If retention is changing, estimate the magnitude (e.g., "recent cohorts retain 8 percentage points better at Month 3 than cohorts from 6 months ago"). - Note any seasonality or external events that may explain shifts. ### Step 4: Recommendations - Based on the patterns, provide 3-5 specific, actionable recommendations. Each recommendation should reference a specific finding from the analysis. - Prioritize recommendations by expected impact on overall retention. - Suggest what additional data would strengthen the analysis (e.g., "segmenting by acquisition channel would clarify whether the Q3 improvement is channel-driven"). ### Step 5: Presentation-Ready Output - Provide the final cohort table in a format I can paste into a spreadsheet or slide. - Write a 3-4 sentence executive summary suitable for a stakeholder email. - Suggest one chart type that would best visualize the key finding.
Usage Tips
- Include at least 4-6 cohorts for meaningful comparison: Two cohorts cannot reveal a trend. Six or more cohorts let the AI distinguish noise from pattern and identify whether retention is improving over time.
- Specify what "retention" means for your product: "Logged in at least once" is very different from "completed a core action." Defining your key metric precisely prevents the analysis from tracking the wrong behavior.
- Segment further after the initial analysis: Once you see the overall cohort table, run the prompt again filtering by acquisition channel, pricing plan, or geography. The biggest insights often come from comparing sub-cohorts within a single time period.
- Set the deviation threshold based on your business maturity: Early-stage products with volatile cohorts should use 15-20 percentage points. Mature products with stable retention should use 5-10 percentage points to catch subtle shifts.
Get more from this prompt
Save it, score it with AI, optimize it, and track every version. Free to start.