Article FAQ

Q: How much data do we need to run cohort analysis?
A: Minimum 12 weeks of customer data per cohort. If you have fewer than 100 customers per cohort, repeat purchase rates are noisy. Aim for 200+ per cohort.

Q: Can we run cohorts on a new store with 2 weeks of data?
A: You can, but results aren't reliable. Retention at week 2 is near 0% for all cohorts. Wait 8-12 weeks to see true repeat purchase patterns.

Q: Should we cohort by customer lifetime or by calendar month?
A: Calendar month is easier operationally. Cohort by the month customers were acquired, then track week-by-week. This ties acquisitions to marketing campaigns and seasonality.

Q: What's a "good" repeat purchase rate?
A: 25-35% at 4 weeks is solid. 40%+ is excellent. Below 20% means retention issues. For high-ticket items ($500+), 15-20% is acceptable; repeat purchases are infrequent.

Q: How do we handle cohorts with seasonal products?
A: Track year-over-year cohorts (Jan 2025 vs. Jan 2024). Don't compare Jan to June—they're different seasons. Use seasonal benchmarks, not month-to-month.