Choosing The Right Retention Window For A DTC Store
Picking 30, 60, or 90 days for retention isn't a stylistic choice — it decides whether your number is honest. Here's how to align the window to your repeat-purchase cycle.
Quick answer
Set your retention window to roughly 1.5× your category's median repeat-purchase interval. For coffee and skincare that lands at 30-60 days; for apparel 90-120; for home and considered-purchase categories 180+. Reporting 30-day retention on a category where customers buy every four months guarantees a number that looks like churn — even when the store is healthy.
Retention window (DTC)
The lookback period over which a DTC store measures whether a customer came back and purchased again.
A retention window is the fixed time span — commonly 30, 60, 90, or 180 days — used to decide whether a customer counts as retained after their first order. The choice looks administrative but it isn't: shortening the window flatters categories with fast repeat cycles (coffee, supplements, pet food) and buries categories with slower ones (apparel, home, electronics). The right window matches how often your customers actually buy, not how often your dashboard defaults refresh.
On Shopify's default reports and most email tools, the window is preset and rarely revisited. That default is where most misreported retention numbers start.
Retention is a ratio, and the denominator is time. Change the time span and you change the story — sometimes by 3× on the same cohort. This page walks through why the window distorts the number, how to spot a mismatch, and how to pick one that survives a board meeting.
Why the window distorts the number
Every product category has a natural repeat rhythm. Ground coffee gets reordered every 3-4 weeks. A moisturiser lasts 6-10 weeks. A pair of running shoes might not come up again for 8 months. If your window is shorter than that rhythm, you're measuring "did they buy again before they'd realistically need to" — which is a different question.
This is what makes 30-day retention flatter consumables and bury considered-purchase brands. A coffee subscription showing 45% 30-day retention looks strong; a denim label showing 8% looks broken. Both stores could be equally healthy — the denim brand's customers simply weren't due back yet.
The comparison trap
Do not benchmark your retention rate against another brand's headline number unless you know the window. "25% retention" is meaningless without "…measured over 90 days on the January cohort." Most public benchmark decks quietly compare 30-day figures to 180-day figures and call it a peer group.
How to detect a window mismatch
Pull the median days-between-orders for repeat customers over the last 12 months. Ignore the mean — one wholesale account can drag it 40 days in either direction. If your window is more than ~50% shorter than that median, you're underreporting retention. If it's more than 3× longer, you're masking churn behind a generous lookback.
A second tell: your Klaviyo 60-day retention number doesn't match the Shopify report. Nine times out of ten that's the two tools using different windows or different anchor dates — not a data bug. Reconcile the windows before you reconcile the numbers.
How to pick a window that holds up
Start with the 1.5× median-repeat-interval rule as your baseline, then cross-check it against your CAC payback period. If it takes you 110 days to pay back acquisition cost on a new customer, a 30-day retention window is telling you almost nothing useful about whether that CAC was worth spending. Matching the window to payback gives you an honest LTV:CAC ratio instead of an optimistic one.
The table below covers the defaults we see hold up across the most common DTC categories. Treat them as starting points — verify against your own median repeat interval before you publish the number anywhere.
Recommended retention windows by DTC category
| Category | Median repeat interval | Recommended window | Why |
|---|---|---|---|
| Coffee / tea | 25-35 days | 45-60 days | Fast consumable; short window still captures the second order |
| Skincare / haircare | 45-70 days | 60-90 days | Product lasts 6-10 weeks; 30-day is too tight |
| Supplements / vitamins | 30-45 days | 45-60 days | Monthly bottles; align with refill cycle |
| Pet food / treats | 30-40 days | 45-60 days | Weight-based reorder cycle |
| Apparel / footwear | 90-150 days | 180 days | Seasonal + wardrobe replacement; 30/60 days buries the story |
| Home / kitchen | 150-240 days | 180-365 days | Considered purchase; retention plays out over quarters |
| Beauty (colour / makeup) | 60-90 days | 90-120 days | Discovery-heavy; second order takes time |
| Electronics / accessories | 180-300 days | 270-365 days | Long replacement cycle; annual view is honest |
Report side-by-side, not single-number
The strongest internal reporting habit is showing 30/60/90/180-day retention on the same chart for every cohort. It removes the argument about "which window is right" and makes the shape of the curve — which is what actually matters — the thing people look at.
Experiment and reporting ideas
Re-run the last four quarters of retention using your new window and check whether the trend line changes direction. If Q3 looked flat on 30-day and rising on 90-day, the window was hiding a real improvement — usually driven by a category or campaign whose payoff sits outside the short lookback.
Then decide between rolling and fixed-cohort reporting. Rolling windows are better for weekly ops reviews; fixed cohorts are better for board decks and post-campaign attribution. Most stores need both, keyed off the same window length so the numbers reconcile.
Frequently asked questions
Neither is universally better — it depends on your repeat-purchase cycle. Use 30-60 days for consumables (coffee, skincare, supplements) and 90-180 days for considered-purchase categories (apparel, home). If your median repeat interval is 80 days, a 30-day window will systematically understate retention.
Calculate the median days-between-orders for your repeat customers over the last 12 months, then multiply by roughly 1.5. Cross-check that number against your CAC payback period — the window should be at least as long as payback for LTV:CAC to be honest.
No. Subscription stores can use tighter windows (often 30-45 days) because the billing cycle enforces a natural rhythm and churn shows up quickly. One-time-purchase stores need longer windows because the repeat decision is discretionary and takes longer to surface.
Almost always a window or anchor-date mismatch, not a data issue. Klaviyo often defaults to 60 or 90 days from the profile creation date; Shopify's cohort report anchors on first order date. Align the window and the anchor before troubleshooting the underlying numbers.
Yes — seasonal categories (swimwear, holiday gifting, back-to-school) break short windows because the repeat purchase is deliberately 6-12 months out. For seasonal SKUs, 90-day retention will look catastrophic; use annualised windows or year-over-year cohort views instead.
Rolling windows suit ops dashboards and weekly reviews because they always show the freshest data. Fixed cohorts are better for board reporting, post-campaign analysis, and LTV modelling because they let you compare cohort to cohort cleanly. Most stores end up running both.
Yes, but publish both versions side by side for at least one reporting cycle so stakeholders can see the shift is a methodology change, not a performance drop. Switching quietly is the fastest way to lose credibility with a finance team.
Directly. A shorter window truncates the revenue captured in LTV, dragging the ratio down; a longer window captures more but risks including revenue you haven't earned yet. Match the window to CAC payback period for the most defensible ratio.
Start with the category default from the table above and mark the number as provisional. Recompute using your actual median repeat interval as soon as you have 3-4 months of second-order data — most stores end up shortening or extending their initial guess by 30-60 days.
If product lines have materially different repeat cycles — for example a coffee brand that also sells brewing equipment — yes. Report a blended figure at the store level plus category-level breakouts. A single window across mismatched categories will always flatter one side and hurt the other.
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.