Retention Funnels
A retention funnel tracks how each customer cohort decays over Day 1, 7, 30, and 90 — and the shape of that curve predicts long-term LTV more reliably than your initial conversion rate.
Retention Funnels
A retention funnel tracks how a customer cohort decays at Day 1, 7, 30, and 90 — the curve shape predicts LTV.
A retention funnel is a time-based view of a single customer cohort — everyone who bought, signed up, or installed in the same window — measured at fixed checkpoints (typically Day 1, Day 7, Day 30, Day 90, and sometimes Day 180). At each checkpoint you record what percentage of the original cohort is still active by whatever signal matters to your store: a repeat visit, a second order, an email open, an app session.
Unlike an acquisition funnel, which is one-shot and ends at checkout, a retention funnel keeps measuring the same people forward in time. The slope of the curve — and crucially, whether it flattens after the initial drop — is what separates a brand with healthy LTV from one that has to keep buying its way to growth.
Most stores obsess over the acquisition funnel — sessions, add-to-carts, checkouts — because that data is loud and immediate. Retention is quieter and shows up weeks later, which is why teams under-invest in it even when CAC is climbing.
The shape of the curve matters more than any single retention number. A cohort that drops to 18% by Day 30 but holds flat through Day 90 is far more profitable than one that hits 25% at Day 30 and then bleeds to 8% by Day 90 — the second cohort looked better and is worth less.
Retention(d) = Active_users_in_cohort_on_day_d / Cohort_size_at_day_0
Retention(d)
Retention rate at day d
Percentage of the original cohort still active on day d
Active_users_in_cohort_on_day_d
Active cohort members on day d
Number of original cohort members who met your activity definition (visit, order, session) on or before day d
Cohort_size_at_day_0
Initial cohort size
Total customers acquired in the cohort window (e.g. all first-time buyers in week of June 3)
A skincare brand acquires 2,400 first-time buyers in the week of a Meta campaign. They measure 'retained' as 'placed a second order or visited the site' within each window.
Cohort size (Day 0): 2,400
Active by Day 7: 1,080
Active by Day 30: 624
Active by Day 90: 504
→ D7 = 45%, D30 = 26%, D90 = 21%
The 5-point drop from D30 to D90 is the healthy signal — the curve is flattening, which means the remaining ~20% are looking like loyal repeat buyers and LTV projections for this cohort can be modelled with more confidence.
Benchmarks vary heavily by category. Replenishment-driven verticals like skincare and supplements naturally retain better than considered-purchase categories like furniture, because the next purchase occasion is closer. Use the table below as a directional reference, not a target — your own historical cohorts are the better baseline.
Typical Day 7 / Day 30 / Day 90 retention for first-time DTC buyers, by vertical
| Vertical | Day 7 | Day 30 | Day 90 |
|---|---|---|---|
| Skincare & beauty (replenishable) | 38-48% | 22-30% | 16-22% |
| Supplements & wellness | 35-45% | 24-32% | 18-25% |
| Apparel (mid-AOV) | 22-30% | 12-18% | 7-11% |
| Home & furniture (high-AOV) | 12-18% | 5-9% | 3-5% |
| Consumer electronics | 18-26% | 8-13% | 4-7% |
| Pet food & accessories | 40-50% | 28-36% | 22-28% |
Retention funnels sit inside the broader practice of funnel analytics — but where acquisition funnels diagnose where you lose visitors in a single session, retention funnels diagnose whether the customers you do win are worth what you paid for them. The two views are complementary: fix the acquisition funnel for volume, fix the retention funnel for profit.
Frequently asked questions
An acquisition funnel measures one-shot conversion within a single session — landing page to product to cart to checkout. A retention funnel measures the same cohort of customers forward in time, checking how many come back at Day 7, 30, 90. Acquisition funnels diagnose conversion friction; retention funnels diagnose product-market fit and LTV.
They roughly correspond to behavioural phases: Day 1 captures immediate post-purchase engagement (order tracking, unboxing), Day 7 catches the first natural reorder consideration for fast-cycle products, Day 30 marks the typical replenishment window for consumables, and Day 90 separates true repeat buyers from one-off shoppers. The specific days are a convention — adjust to your purchase cycle.
Churn rate is usually a single number (e.g. '6% monthly churn') applied to a whole base. A retention funnel is cohort-specific and time-resolved — it shows you that the cohort acquired during a Black Friday sale churns twice as fast as the cohort from a brand campaign, which a blended churn number hides entirely.
Yes. Two cohorts can hit identical Day 30 numbers and have wildly different lifetime values depending on whether the curve flattens or keeps falling. A flattening curve means the survivors are loyal and predictable; a continuing decline means you're still losing your 'best' customers and any LTV projection based on D30 alone will overstate revenue.
GA4 has a Cohort Exploration report that lets you define an acquisition event (first_purchase) and a return event (any session, purchase, or custom event), then bucket users by acquisition week and measure retention at fixed intervals. It works but it's slow and limited to event-level data — most teams export to a warehouse or use a dedicated analytics tool for richer cohort segmentation.
It depends entirely on category. Replenishable categories (beauty, supplements, pet) typically hit 22-32% at Day 30; considered purchases (furniture, electronics) come in at 5-15%. The more useful comparison is against your own past cohorts — a downward trend across recent months is a louder signal than any industry number.
Aim for at least 200-300 customers per cohort for stable percentages — below that, single-customer fluctuations move the curve too much to read. If your weekly volume is lower, group into monthly cohorts. The trade-off is responsiveness: monthly cohorts smooth out noise but you wait 30+ days to see how a campaign affected retention.
LTV is essentially the area under the retention curve multiplied by average order value and reorder frequency. A more retentive curve directly translates to a higher LTV for the same acquisition cost. This is why teams running predictive LTV models almost always start by fitting a curve (typically power-law or exponential) to historical retention funnel data.
Common culprits: a product experience problem (the item arrived, was used once, didn't impress), a poor onboarding sequence (no Day 3-10 lifecycle email), or a misaligned acquisition source (discount-driven traffic that was never going to repeat). Compare the curve across acquisition channels to isolate which cause is yours.
Almost always yes. Paid social, paid search, organic, email, and referral cohorts retain very differently — referral and organic typically retain 1.5-2x better than discount-driven paid social. Without segmentation, a shift in your channel mix can make your blended retention look like it's collapsing when individual channels are stable.
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.