90-Day RPR vs 12-Month Cohort Retention For Consumables
A category-specific playbook for consumables: when a 90-day RPR is the honest health metric, when only a 12-month cohort curve tells the truth, and how to size the replenishment window for skincare, supplements, and coffee.
Quick answer
Use 90-day RPR as your primary health metric only if your median replenishment cycle is under 60 days (most coffee, many supplements). For skincare and any SKU with a 60-120 day cycle, 90-day RPR flatters new cohorts — pair it with a 12-month cohort curve, and read the second-purchase peak, not the raw rate.
90-Day RPR vs 12-Month Cohort Retention for Consumables
A category-specific rule for choosing between 90-day repeat purchase rate and 12-month cohort retention when the SKU has a natural replenishment cycle.
Consumables — skincare, supplements, coffee, pet food, cleaning refills — burn down at a predictable rate. That physical cycle is what makes retention measurable, but it's also what makes short windows deceptive: a 90-day repeat purchase rate captures a fast-cycle SKU cleanly and misses a slow-cycle one entirely.
The decision comes down to sizing the measurement window against the replenishment window. If 80% of a first-order cohort should physically run out inside your window, 90-day RPR is honest. If they haven't, you need a 12-month cohort curve to see the truth — and to avoid rewarding acquisition channels that only look good for a quarter.
This is a sub-decision under the broader repeat purchase rate vs retention rate question. The general framing tells you which family of metric fits your business model; this page tells you which specific window fits your specific SKU chemistry.
The mistake most Shopify skincare and supplement brands make: they picked 90-day RPR because it was the default in a dashboard tool, and it's been quietly overstating retention on their 90-day serum for two years.
Sizing the replenishment window
Start with the median days-to-second-purchase for repeat buyers on the specific SKU — not the brand average. A 30ml serum and a 500ml cleanser from the same brand can have wildly different replenishment cycles.
Then apply the 1.5× rule: your measurement window should be at least 1.5× the median replenishment cycle. Ground coffee at 21 days → 32-day window minimum. A supplement 30-count bottle at 30 days → 45 days. Retinol serum at 75 days → 113 days, so 90-day RPR is already tight.
The flattery zone
If your median replenishment cycle is 60-120 days and you report 90-day RPR, you're measuring the fastest ~40% of your cohort and calling it retention. New customers who bought a bottle in month one and are still using it in month three show up as churned. That's not a health metric — that's a vanity metric with a survivorship bias.
When 90-day RPR is the honest metric
Fast-cycle consumables where the pack size and dose rate mean physical burn-down inside 45-50 days. Whole-bean coffee at one bag per week, daily-use supplements in 30-count bottles, single-serve pods, cleansers under 200ml at 2x/day use.
In these cases, 90-day RPR is directionally correct and fast enough to be operationally useful. You can read it monthly, react to acquisition channel shifts inside a quarter, and use it as the KPI for a replenishment email flow without waiting a year for signal.
The tell: your second-purchase distribution peaks between day 25 and day 60, and 90-day and 120-day RPR are within 5 percentage points of each other. When those two windows converge, the short window is telling the truth.
Replenishment cycles by consumable subcategory
Typical replenishment cycles and honest measurement windows by consumable subcategory
| Subcategory | Median days to 2nd order | Recommended window | 90-day RPR reliable? |
|---|---|---|---|
| Ground/whole-bean coffee (subscription-eligible) | 21-28 | 45 days | Yes |
| Daily supplements (30-count) | 28-35 | 60 days | Yes |
| Daily supplements (60-count) | 55-70 | 120 days | Borderline |
| Cleansers & shampoo (<250ml) | 45-60 | 90 days | Borderline |
| Serums & actives (30ml) | 70-95 | 150 days | No — use cohort curve |
| Moisturisers (50ml) | 60-80 | 120 days | No — use cohort curve |
| Retinol/prescription-adjacent (15-30ml) | 90-130 | 180 days | No — 12-month cohort only |
| Pet food (medium bag) | 30-45 | 75 days | Yes |
These are Shopify-band ballparks for €1M-€15M brands. Your own numbers will shift with pack-size mix, subscription penetration, and whether you sell into Shopify Markets regions with different usage norms — pull the SKU-level median before trusting any category default.
When only a 12-month cohort curve tells the truth
Any SKU where the replenishment cycle is 60+ days: serums, moisturisers, retinol, larger supplement bottles, premium coffee bought in bulk. Here the second-purchase peak sits at month 3-4, and a 90-day window catches maybe half of the eventual repeaters.
A 12-month cohort curve fixes two things at once. It shows the true second-purchase peak (so you know when to trigger replenishment flows), and it lets you compare cohorts by acquisition channel across a full annual cycle — catching the paid-social cohorts that convert fast and churn faster than they should.
How to run both without doubling the work
Report 90-day RPR weekly as the leading indicator, and refresh the 12-month cohort curve monthly as the truth check. The RPR moves fast enough to catch acquisition-quality shifts; the cohort curve confirms whether those shifts are real or just replenishment-timing artefacts.
Historical GA4 imports make this practical from day one — you don't need to wait 12 months to build the first cohort curve, you backfill it. Then anchor your replenishment email flow, subscription upsell timing, and paid-channel LTV assumptions to the cohort peak, not the 90-day snapshot.
Experiment ideas
Test replenishment email timing against your actual cohort peak: if the second-purchase distribution peaks at day 78 for your 50ml moisturiser, run send-at-day-60 vs send-at-day-70 vs send-at-day-80. Most brands are sending 2-3 weeks too early because they anchored to the 90-day window.
Second test: pack-size framing at PDP. Offering a 100ml alongside the 50ml doesn't just lift AOV — it collapses the replenishment cycle uncertainty, which makes 90-day RPR reliable again. Measure the shift in both metrics on the cohort that took the upsize.
Frequently asked questions
Because you can't wait a year to react. 12-month cohort curves are the truth, but they're slow. 90-day RPR gives you a reading you can act on inside a quarter — as long as your replenishment cycle actually fits inside the window. Use both, at different cadences.
Report 90-day RPR at the SKU or category level, not brand-wide. A brand-wide 90-day RPR that blends coffee (28-day cycle) with retinol (110-day cycle) is meaningless — the fast SKU carries the number and hides the slow SKU's leakage.
Subscriptions make 90-day RPR look great by construction — the second order is baked in. Report subscription and one-time cohorts separately, and hold the 12-month cohort curve as the honest metric for the one-time buyers, who are the real acquisition test.
It's a rule of thumb: at 1.5× the median cycle, roughly 70-80% of a cohort has physically run out and had the opportunity to repurchase. Below that, you're measuring intent-to-repurchase in customers who haven't finished the product yet, which isn't retention.
Yes — pet food, refillable cleaning products, contact lenses, protein powder, and vape liquid all follow the same logic. Any SKU where physical burn-down drives repurchase timing is subject to the replenishment-window sizing rule.
Filter orders for a first-purchase cohort (say, all customers whose first order was in Q1), then pull each customer's second order date. The median gap is your replenishment cycle for that SKU. Do this at the SKU or line-item level, not the customer level, or you'll blend cycles.
No — they're related but distinct, which is covered in the parent repeat purchase rate vs retention rate framing. RPR is transaction-based (did they buy again?); retention rate is customer-based (are they still active?). For consumables, RPR tends to be the more operational metric because it maps directly to replenishment behaviour.
Larger pack sizes lengthen the replenishment cycle and push you toward needing the 12-month cohort curve. A 60-count supplement bottle behaves like a serum, not a supplement. Segment your RPR by pack size — the two windows are different products from a measurement standpoint.
Anchor the send to your actual second-purchase peak from the cohort curve, then test ±10 days around it. For a 90ml retinol with a 110-day peak, that means a primary send at day 95, a reminder at day 115, and a win-back at day 150 — not the generic 60/90/120 template.
Yes, for the 60-120 day cycle band it's often a good middle ground: fast enough to be operational, wide enough to catch the real second-purchase peak. It's underused because most tools default to 30/60/90/365, but 180 is the honest window for a lot of skincare.
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.