.svg){kind=small}
TL;DR
- Eight retention metrics matter for ecommerce: repeat purchase rate, second-order probability, purchase frequency, cohort retention, churn, AOV, LTV, and net revenue retention.
- Measure every one of them per cohort, not blended. Blended numbers hide channel mix drift and kill the early warning.
- A healthy D2C brand has a repeat purchase rate of 25–40%. Best-in-class categories hit 45–60%.
- Net revenue retention over 100% means cohort revenue grows without new customers. Under 80% and the business depends entirely on paid acquisition.
- Fairview joins Shopify, Stripe, and ad data, then surfaces the eight metrics per cohort and channel automatically.
Every operator meeting at a growth-stage ecommerce brand starts with acquisition numbers. CAC is down, ROAS is up, sessions are flat. Two hours later, half the room realizes the business is quietly running on a 14% repeat purchase rate and the cash does not add up. Retention is the half of the story that matters, and most reporting tools do not show it.
This piece is the retention counterpart to profit leak detection. Eight metrics worth tracking, the formulas, benchmarks by category, and a quick framework for which to watch weekly vs monthly vs quarterly. Paired with contribution margin by channel, it covers both sides of the unit economics picture.
Why retention, measured right, catches problems early
Definition
Customer retention (ecommerce): the percentage of customers from a given acquisition window who place another order within a defined time frame. Always measured per cohort. Blended retention numbers are nearly useless because they average channel effects that move independently.
The case for retention is simple. Acquiring a new customer costs 5–7x more than selling to an existing one, per the often-quoted Bain research. Whether that multiplier is exactly right matters less than the direction: every incremental repeat purchase subsidizes the CAC on the first order, and most D2C brands that died quietly did so because the subsidy never showed up.
Retention is also the earliest honest signal of product-market fit. A cohort that does not come back was not really served by the first purchase, no matter how strong the acquisition numbers looked that week.
The eight retention metrics that matter
1. Repeat purchase rate (RPR)
The share of customers from a window who place at least one more order within a defined period. Formula: customers with 2+ orders / total customers. The most honest top-line retention number an ecommerce brand has. Healthy D2C: 25–40%. Best-in-class (coffee, supplements, beauty): 45–60%. Under 20% means the business lives on paid acquisition.
2. Second-order probability
The probability a first-time buyer places a second order, typically within 90 days. Sharper than RPR because it isolates the first-to-second transition, which is where 60–70% of lifetime value is determined. If this number is improving but RPR is flat, long-term LTV is quietly rising before the monthly numbers reflect it.
3. Purchase frequency
Orders per customer per period. Formula: total orders in period / unique customers in period. Watch the trend per cohort. A rising frequency with flat AOV is a reorder flywheel finding its rhythm. A falling frequency with rising AOV is a brand losing the core customer and picking up a few heavier buyers — which is usually unsustainable.
4. Cohort retention curve
The percentage of a monthly cohort still active at month 1, 3, 6, 9, 12. This is the single most informative retention view for ecommerce. Plot cohorts side by side and look for the curve that slopes downward over time — recent cohorts retaining worse than older cohorts is the earliest signal of quality drift, promo tax, or channel mix shift.
5. Churn rate
For subscription models: the share of active subscribers who cancel in a period. For non-subscription ecommerce: a definition choice you have to make, usually "customers who have not purchased in 180 days divided by customers active 180 days ago." Always document the definition, because churn is the most misquoted retention metric in D2C.
6. Average order value (AOV)
Revenue per order. Not strictly a retention metric but critical context. A rising AOV from repeat customers is a sign of deepening loyalty. A rising AOV driven by one-time promo bundles is a sign the next-order value is falling. Always segment first-order AOV from repeat-order AOV.
7. Lifetime value (LTV)
Total contribution margin a cohort generates over its observed lifetime. The output of retention, frequency, AOV, and contribution margin combined. Track LTV at 3, 6, 12, and 24 months. Pair with CAC payback to understand whether growth is compounding or burning cash.
8. Net revenue retention (NRR)
Revenue from an existing cohort over time, including repeat orders and subscription upgrades, minus refunds and churn. Expressed as a percentage of the starting revenue. Over 100% means the cohort is growing without new customers. Under 80% and the business is on a treadmill — every dollar of new revenue has to replace a dollar of lost revenue before growth shows up.
Key insight
Every retention metric has to be measured per cohort. A blended retention number is an average of averages, and averages hide every interesting problem in the business.
Benchmarks by category
| Category | RPR (12mo) | 2nd Order 90D | NRR (12mo) |
|---|---|---|---|
| Beauty / skincare | 40–55% | 30–45% | 95–115% |
| Supplements / wellness | 45–60% | 35–50% | 100–120% |
| Coffee / consumables | 50–65% | 40–55% | 110–130% |
| Apparel | 20–32% | 12–20% | 80–100% |
| Home goods / one-time | 15–28% | 8–14% | 60–85% |
Category matters enormously. Holding an apparel brand to a coffee-brand RPR is the fastest way to start a pointless planning meeting. Pick the band that matches your category, then watch the trend within your own business for the real story.
The cadence that catches churn early
A workable cadence that keeps retention on the operating table without turning it into a research project:
- Weekly. Repeat purchase rate and second-order probability for the most recent four cohorts. Any drop of 3+ points triggers a look at channel mix and promo activity.
- Monthly. Cohort retention curves for the last six months of acquisitions. 90-day LTV, AOV split by first vs repeat order, frequency per cohort.
- Quarterly. Net revenue retention, 12-month LTV, category benchmarks comparison, and any churn-definition changes. This is the meeting where retention feeds the next quarter’s acquisition mix decisions.
Quote-ready
The earliest signal that an ecommerce brand is in trouble is almost always a drop in second-order probability three months before revenue notices.
Three retention mistakes that cost money
- Reporting blended instead of cohort. A blended 34% repeat purchase rate can be 48% organic and 19% paid social. When the paid-social share grew, the blended fell. Nothing else changed, and every slide in the deck pointed the wrong direction.
- Confusing AOV growth with retention growth. A brand pushing bundled promos sees AOV rise and retention flat. The top-line looks healthy. The repeat rate is quietly softening because promo-acquired cohorts reorder less.
- Using booked revenue instead of collected. NRR computed from gross revenue ignores refunds and chargebacks. A 105% NRR on gross is often 92% on net. Track net.
How Fairview tracks ecommerce retention automatically
Fairview connects to Shopify, Stripe, QuickBooks, Xero, HubSpot, Salesforce, Pipedrive, Google Ads, Meta Ads, and HubSpot Marketing Hub via native OAuth. Once connected, the operating view reconstructs monthly cohorts, enriches with channel attribution, and reports the eight retention metrics per cohort and per channel without a warehouse project.
When a cohort’s second-order probability drops past a configured threshold, Fairview writes a named next-best action: "April cohort second-order probability fell from 32% to 21% vs the trailing three cohorts. Driver: paid-social share rose from 38% to 54% of acquisitions. Recommendation: re-check AOV on first order by channel and tighten paid-social creative."
See pricing and tiers for the plan that fits your stack.
8 metrics
Reported per cohort, not blended
Weekly
RPR + 2nd order refresh
10
Native integrations live today
Key takeaways
- Track eight retention metrics: RPR, second-order probability, frequency, cohort retention, churn, AOV, LTV, NRR.
- Measure per cohort. Blended retention hides every interesting problem.
- Benchmark by category. Coffee and apparel are not the same business.
- Cadence: weekly RPR and 2nd-order, monthly cohort curves, quarterly NRR.
- Second-order probability drops are the earliest honest signal of trouble ahead.
See the eight retention metrics per cohort.
Connect Shopify, Stripe, and your ad accounts. Fairview returns the eight retention metrics per cohort and channel on day one. 14-day trial, no card required.
Frequently asked questions
Repeat purchase rate, second-order probability, purchase frequency, cohort retention curves, churn rate, AOV split by first vs repeat order, lifetime value, and net revenue retention. All eight should be measured per monthly cohort, not as a blended number, because blended averages hide channel mix shifts that move retention independently.
Healthy D2C brands see a 12-month repeat purchase rate of 25 to 40 percent. Best-in-class categories like coffee, supplements, and beauty land at 45 to 60 percent. Below 20 percent and the business is structurally dependent on continuous paid acquisition, which gets expensive fast as CAC rises.
Customer retention rate equals the number of customers at the end of the period, minus new customers acquired in the period, divided by customers at the start of the period, expressed as a percentage. For ecommerce, run this per monthly acquisition cohort rather than at a total-customer level so channel mix and promo effects become visible.
Net revenue retention tracks revenue from an existing cohort over a given period, including repeat purchases, subscription upgrades, and product add-ons, minus churn, cancellations, and refunds. Expressed as a percentage of the starting revenue for that cohort. NRR above 100 percent means the cohort is growing without new customers; under 80 percent means the business depends entirely on new acquisition to grow.
Weekly for the fastest-moving metrics: repeat purchase rate and second-order probability on recent cohorts. Monthly for cohort retention curves, 90-day LTV, and frequency trends. Quarterly for net revenue retention, 12-month LTV, and category-benchmark comparisons. The weekly cadence is where problems get caught before they compound.
Blended retention is an average across every cohort and channel. If paid-social cohorts retain at 15 percent and organic at 48 percent, a shift in the acquisition mix toward paid social drops the blended number without anything else changing in the business. Cohort-level reporting exposes the cause. Blended numbers hide it.