30 Day Vs 90 Day Vs 365 Day Retention Windows: When Each One Is Honest
A head-to-head of the three default retention windows — with the category profile each one actually measures honestly, and the ones each one flatters or buries.
30-day vs 90-day vs 365-day retention windows
The three default repeat-purchase measurement windows, each honest for a different purchase-cycle profile.
A retention window is the time horizon over which you measure what share of a cohort came back to buy again. The three defaults — 30, 90, and 365 days — are not interchangeable. Each one tells the truth for one kind of catalogue and lies about the others.
A 30-day window reads honestly for daily-use consumables (coffee, pet food, skincare refills) because the repurchase cycle fits inside it. A 90-day window suits mid-cycle categories like apparel and supplements. A 365-day window is the only honest frame for considered purchases — furniture, appliances, mattresses — where a repeat buyer may take 8-14 months to come back.
The window you pick decides whether your retention curve looks like a success story or a disaster — on the same data. That is why defaulting to whichever window your dashboard shows first is a reporting choice, not a measurement choice.
The honest rule is simple: the window must be at least as long as the median repurchase interval for your category, and short enough that cohort sizes stay statistically readable. Everything below is how that rule plays out across the three defaults.
The decision table: which window is honest for which catalogue
| Window | Honest for | Flatters (overstates) | Buries (understates) | Typical repeat rate seen |
|---|---|---|---|---|
| 30 days | Daily-use consumables: coffee, pet food, vape pods, skincare refills, protein | Subscription-heavy brands with auto-ship | Apparel, supplements, furniture, appliances | 18-35% |
| 90 days | Mid-cycle: apparel, cosmetics, supplements, accessories, home goods under €80 | Seasonal apparel measured off a launch month | Furniture, mattresses, appliances, considered electronics | 22-40% |
| 365 days | Considered purchases: furniture, mattresses, large appliances, premium electronics, bikes | Any category where seasonality cycles once | Nothing — but cohort size often becomes too thin to read on stores under €2M | 15-30% |
Two columns deserve a second look. The "flatters" column is where the window is wide enough to catch a second purchase that was never really a loyalty signal — it was just the next scheduled refill or the next season's drop. The "buries" column is where the window closes before the typical customer has had a chance to come back.
Why the wrong window changes the story on the same cohort
Take a mid-market apparel store with a 110-day median repurchase interval. Measured at 30 days, their January cohort shows a 9% repeat rate — the executive team concludes retention is broken. Measured at 90 days the same cohort shows 24%. At 365 days it shows 41%.
Nothing changed about the customers. The 30-day number is simply measuring a window shorter than the physical shopping cycle for jeans. For a related breakdown of why a 30-day window flatters consumables and buries considered purchases, the mechanism is the same in reverse: the window is lying about what the cohort is doing.
The screenshot trap
If your GA4 default is a 30-day retention report and your Shopify dashboard defaults to 90 days, you will see two different numbers for the same cohort every Monday. Pick ONE window per category, document it, and stop toggling. Toggling is where leadership loses trust in the metric.
The hidden cost of 365-day windows on smaller stores
The 365-day window is the honest choice for furniture or appliances — but only if your cohort is big enough to survive the thinning. A store doing €1M in revenue with a €180 AOV ships roughly 5,500 orders a year, which splits into monthly cohorts of ~460 buyers. By month 12, random noise is comfortably ±4 percentage points on the retention rate.
That means two consecutive quarters can show a 6-point swing with no underlying change in customer behaviour. Operators then chase phantom causes — a creative refresh, a pricing test, a new email flow — when the only signal is cohort thinning. The usual fix is to aggregate cohorts quarterly instead of monthly, or move to a shorter window if your category tolerates it.
Repeat-purchase rate on the same apparel cohort, measured at three windows
Frequently asked questions
90 days is the honest default for most apparel catalogues, because the median repurchase interval sits in the 70-130 day range. If you sell seasonal-only pieces (coats, swim) you can stretch to 180 days, but 365 days overstates loyalty by catching the next season's purchase as a "repeat".
Not wrong, but not the headline number. A 30-day read on apparel or furniture is useful as a leading indicator of post-purchase experience — did they come back fast because they loved it, or did they come back fast to buy a second size because fit was off? Just do not report it as retention.
Split the cohorts by first-purchase category and measure each on its own window. A beauty brand that also sells devices should run 30-day retention on the refill SKUs and 180- or 365-day retention on the device line. One blended number across both will lie about both.
Yes, directly. LTV built off a 30-day retention curve for a furniture brand will understate value by a factor of 2-3x, because it ends before the typical repeat purchase has happened. Match the LTV horizon to the category's purchase cycle, not to the window your dashboard defaults to.
Two reasons. First, cohorts on stores under €2M thin out to statistical noise by month 12. Second, 365 days is too slow a feedback loop for anything you want to optimise — you cannot steer a retention programme on data that lands 12 months late.
A rolling 90-day window (every customer measured 90 days from their own first order) gives you a continuously-updating number. A fixed window (all customers from the January cohort, measured at 90 days) is cleaner for cohort comparisons. Use rolling for operational dashboards and fixed for analysis.
Subscription inflates every window. A 30-day retention number on an auto-ship coffee brand is really measuring billing-cycle completion, not customer choice. Report subscription retention separately and keep a one-time-buyer cohort to see honest loyalty.
No, and benchmarks that mix windows are the main reason "industry retention rates" are so unreliable. Only compare like-for-like windows on like-for-like categories. If a benchmark report does not specify both, treat the number as directional at best.
Longer windows need bigger cohorts to stay readable. Rough rule: aim for at least 300-500 buyers per cohort at the window endpoint. If you cannot hit that at 365 days, aggregate to quarterly cohorts or pick a shorter window your category can still tolerate.
Do not switch quietly — leadership will notice the number jumped. Publish both windows side-by-side for one quarter with a note explaining why the new one is more honest, then retire the old one. The transition cost is one awkward slide; the alternative is years of a misleading KPI.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.