Choosing the Right Churn Window: Monthly, Quarterly, or Cohort
Monthly churn over-reacts to BFCM, quarterly smooths it, and cohort churn is the only honest read when repurchase cycles stretch past 60 days. Here's how to pick.
Quick answer
Use monthly churn only if your median repurchase cycle is under 45 days (consumables, pet food, supplements). Use quarterly for apparel, home, and most beauty. Use cohort churn whenever the repurchase cycle exceeds 90 days or you run heavy promotional spikes like BFCM — anything else inflates or deflates the number in ways that mislead the team.
Choosing the Right Churn Window
Picking the measurement window (monthly, quarterly, or cohort) so your churn number reflects real customer behaviour, not calendar noise.
The churn window is the time period over which you decide a customer has stopped buying. The Churn Rate Calculator gives you the same formula regardless of input — but the window you feed it determines whether the answer is honest or theatrical.
Monthly windows flatter brands with short repurchase cycles and punish brands with long ones. Quarterly windows smooth promotional noise but lag real deterioration. Cohort-based churn tracks a specific group of acquired customers over their natural lifecycle, which is the only read that holds up when repurchase cycles stretch or promotional spikes distort month-over-month comparisons.
Most operators inherit a monthly churn dashboard because that's what GA4 and Shopify default to. Then they ship a misleading number to the leadership deck — usually one that says churn is fine when it isn't, or panics the team in January because December cohorts look thin.
Why monthly windows mislead most stores
Monthly churn assumes a customer who hasn't repurchased in 30 days is gone. For a consumables brand selling 30-day SKUs that's defensible. For an apparel store where the median gap between orders is 95 days, it labels 80% of healthy customers as churned every month.
The second failure mode is promotional distortion. November and December acquire customers at discount, many of whom were never going to repurchase at full price. A January monthly churn read spikes — not because retention got worse, but because the denominator changed shape.
The BFCM trap
A Shopify apparel store sees monthly churn jump from 18% in October to 31% in January and the team launches a panic retention campaign. The real story: BFCM brought in 4,200 discount-only buyers who were never going to come back. Strip them into their own cohort and the underlying churn rate didn't move.
How to detect that your window is wrong
Pull your last 12 months of orders and calculate the median days between first and second purchase. If that number is above 45 days, your monthly churn is structurally inflated. If it's above 90 days, monthly churn is essentially fiction and you should be reading quarterly or cohort numbers.
The second tell: variance. If your monthly churn swings more than 8 percentage points between consecutive months without a clear operational cause, the window is too short for your repurchase cadence. Switch to quarterly and the noise collapses — that's the signal you needed quarterly all along.
How to pick the right window
Start from your repurchase cycle, not from what your dashboard offers. Consumables and subscriptions (pet food, vitamins, coffee) live on monthly. Apparel, beauty refills, and home goods belong on quarterly. Furniture, electronics, and considered-purchase verticals only make sense on cohort or annual reads.
When you run heavy seasonal acquisition — BFCM, Mother's Day, back-to-school — cohort churn is non-negotiable. Tag each acquisition cohort and track its repurchase rate at 30, 60, 90, and 180 days. That separates the discount-only buyers from the customers worth retaining, and stops January from looking like a crisis.
The repurchase-cycle rule
Pick a window that is at least 2× your median repurchase cycle. A 60-day median repurchase means quarterly is the floor. A 120-day median means semi-annual or cohort-based. This single rule eliminates 90% of misleading churn reports.
Experiments to validate your new window
Before you switch the dashboard, back-test. Re-run the last 18 months of orders through the Churn Rate Calculator at monthly, quarterly, and 90-day cohort settings. Plot all three. The window whose curve aligns with known business events (a stockout, a price hike, a hero product launch) is the one telling the truth.
Then run a retention experiment on one cohort — a win-back flow at 75 days for apparel, say — and measure its impact at the cohort's natural repurchase window, not at the next month-end. If the dashboard window can't see the lift, the window is wrong, not the experiment.
Frequently asked questions
Monthly churn measures the percentage of all active customers who didn't buy this month. Cohort churn tracks a specific group of customers acquired in the same period and measures repurchase rate over their natural lifecycle. Cohort is more honest for any store with a repurchase cycle longer than 45 days.
No. Quarterly churn is calculated on a 90-day window with a 90-day lookback, so it captures customers whose natural cadence is 45-90 days. Multiplying monthly churn by 3 overstates the number because it double-counts customers who skipped one month but bought the next.
Pull all customers with 2+ orders from the last 12 months, calculate days between first and second order for each, take the median. Shopify exports and the Churn Rate Calculator's cohort view both surface this. Under 45 days = monthly works; 45-90 = quarterly; over 90 = cohort.
Yes, for any store where BFCM contributes more than 15% of annual orders. BFCM cohorts churn 2-3× faster than full-price cohorts because discount-only buyers self-select out. Folding them into a blended monthly number makes January-February look catastrophic when underlying retention is stable.
Monthly is fine for monthly-billing subscriptions because the window matches the billing cycle. For longer billing cycles (quarterly subscription boxes, annual plans) match the window to the renewal cadence. Anything else creates artificial churn around renewal months.
Report the window that matches your repurchase cycle as the headline, and add cohort retention at 90 and 180 days as supporting context. Three competing churn numbers in a leadership deck causes more confusion than it resolves — pick the honest one and stick to it.
LTV is roughly AOV × purchase frequency × (1 / churn rate). If your churn rate is inflated by the wrong window, LTV is understated, CAC payback looks worse than reality, and paid acquisition gets throttled unnecessarily. Fixing the window often unlocks 20-40% headroom on acquisition spend.
Yes, but backfill the previous 12 months on the new window before changing the dashboard. Otherwise the change itself looks like a churn movement and triggers false alarms. Document the switch in the dashboard so future readers know why the trend line discontinues.
Separate them. First-purchase churn (probability of a second order) behaves very differently from returning-customer churn and benefits from cohort treatment. Returning customers with 3+ orders are usually stable enough that quarterly or even monthly reads work.
Yes — the window input on the Churn Rate Calculator accepts monthly, quarterly, and cohort-based settings, and switching between them on the same input data is the fastest way to see how much your window choice was distorting the answer.
Test ideas before you ship them
Run unlimited A/B tests, attach hypotheses to outcomes, and build a searchable archive of what works — and what doesn't.