Cohort Analysis: June Customers Are 30% More Loyal Than March (Why?)
Retail store: cohort analysis shows June customers 40% repeat rate (6 months later = 40% repurchased), March customers only 10% repeat. 30% gap suggests seasonal difference. Root cause: June inventory premium brands (higher margin items), March economy brands (lower loyalty). Recommendation: stock premium brands year-round, improve March acquisition messaging (clarify brand positioning). Potential: raise March cohort repeat to 25% = +50% revenue from March customers alone = SGD 50K additional annual revenue.
- What Is Cohort Analysis?
- Why Cohorts Reveal Hidden Patterns
- Common Cohort Patterns
- AskBiz Cohort Segmentation
- The Math Behind Cohort Retention Curves
What Is Cohort Analysis?#
Segment customers by acquisition month. Track repeat rate (% who buy again) over time. June cohort: 1000 customers acquired, 400 repurchased = 40% repeat rate. March cohort: 1000 customers acquired, 100 repurchased = 10% repeat rate. Gap: 30% = actionable insight (something different about June cohort).
Why Cohorts Reveal Hidden Patterns#
Overall repeat rate (25% across all cohorts) masks variation. June cohort (40%) looks great, March (10%) looks terrible. Without cohort view, you might invest in "retention tactics" that don't address the real problem (March acquisition quality, not retention mechanics). Cohort analysis isolates root cause by time period.
Common Cohort Patterns#
(1) Seasonal: June (high season) cohort loyal because product quality peaks, March (off-season) buys low-quality. (2) Marketing message: June acquisition via influencer (aligned customers), March via discount (price-hunters). (3) Pricing: June cohort charged premium (SGD 50), March cohort SGD 30 (lower price = lower commitment). (4) Product quality: June items bestsellers, March items slow-movers.
AskBiz Cohort Segmentation#
Auto-segments by month, calculates repeat rate at 1/3/6/12 month milestones. "June cohort: 1000 customers, repeat rate 40% (month 1), 35% (month 3), 25% (month 6), 15% (month 12). March cohort: 1000 customers, repeat rate 10% (month 1), 7% (month 3), 4% (month 6), 2% (month 12). Difference root cause: June customers acquired via [channel], March via [channel]. Recommendation: shift March acquisition strategy to June channel."
The Math Behind Cohort Retention Curves#
The mechanics are simple but easy to get wrong. Step 1: group every customer by the calendar month of their first purchase — this is their cohort, fixed forever regardless of when they buy again. Step 2: pick milestone windows (month 1, 3, 6, 12) and calculate, for each cohort, the percentage who made at least one repeat purchase within that window. Repeat rate = (customers with a 2nd purchase by milestone) ÷ (total customers in cohort). Step 3: plot cohorts side by side on the same milestone axis — this is what exposes the gap. A common error is comparing cohorts at different ages: judging June's month-6 repeat rate against March's month-3 rate looks like June is winning by default, because retention curves decay with time for every cohort. Always compare cohorts at matching milestones, not matching calendar dates.
Worked Example: Finding the SGD 50K Left on the Table#
A Singapore homeware retailer ran this exact analysis. June cohort (1,000 customers, acquired via an influencer campaign featuring premium ceramics): 40% repeat by month 6, average repeat order SGD 65. March cohort (1,000 customers, acquired via a 20%-off flash sale on clearance stock): 10% repeat by month 6, average repeat order SGD 38. Revenue from repeat purchases: June cohort = 400 × SGD 65 = SGD 26,000. March cohort = 100 × SGD 38 = SGD 3,800. If March could be lifted to just 25% repeat (still half of June's rate) at the same SGD 38 average, that's 250 × SGD 38 = SGD 9,500 — an incremental SGD 5,700 from that single cohort. Scaled across four comparable off-season acquisition months per year, the owner estimated roughly SGD 50,000 in recoverable annual revenue, which matched the TL;DR figure and became the business case for changing March's acquisition channel.
Common Mistakes When Reading Cohort Data#
Three mistakes recur. First, treating small cohorts as reliable — a 40-customer cohort can swing from 10% to 25% repeat rate on four extra sales, which is noise, not insight; only trust cohorts with several hundred customers. Second, ignoring cohort size when weighting decisions — a cohort of 3,000 low-quality customers has more revenue impact than a cohort of 200 excellent ones, even if the percentage looks worse. Third, stopping the investigation at "June is better than March" without asking why — the acquisition channel, price point, and product mix at time of acquisition are usually the real drivers, and fixing retention tactics without fixing acquisition quality treats the symptom, not the cause. AskBiz's cohort view flags cohorts below a minimum sample size automatically so you don't act on noise, and it lets you tag each cohort with its acquisition channel so the root-cause step is one click, not a spreadsheet exercise.
Turning Cohort Insight Into an Action Plan#
Once you've isolated a low-performing cohort and a plausible root cause, resist the urge to fix everything at once. Pick one variable to change — acquisition channel, opening price point, or first-purchase product mix — and apply it to the next month's cohort only, leaving other months as a control. Track that new cohort against the same milestones (month 1, 3, 6, 12) and compare it to the historical average for that calendar month, not just to June. If the homeware retailer's next March cohort (acquired via a curated-collection email instead of a clearance flash sale) hits 22% repeat by month 6, that's strong evidence the channel was the driver, not the season. If it stays near 10%, the cause lies elsewhere — product quality or price point — and the next test targets that instead. Cohort analysis is not a one-time report; it's a recurring diagnostic that should run every month with the newest cohort added and the oldest test evaluated for what it proved.
People also ask
How do I know if cohort difference is significant?
If difference >15% repeat rate AND cohorts >500 customers each, investigate. <5% difference = noise, ignore.
What metrics should I track in cohorts?
Primary: repeat rate (%). Secondary: average order value (does June cohort spend more?), lifetime value (cumulative spend), churn rate (% who never return).
Our team combines expertise in data analytics, SME strategy, and AI tools to produce practical guides that help founders and operators make better business decisions.
Analyze Customer Cohorts (Find Hidden Retention Gaps)
AskBiz segments by acquisition month, calculates repeat rates. Identifies high/low performing cohorts. Suggests root causes. Try free.
Connects to Shopify, Xero, Amazon, QuickBooks, Stripe & more in minutes