E-commerce subscription churn cost: replenishment vs curation models

Replenishment vs curation: the structural split

Replenishment subscriptions solve a recurring need: pet food, razor blades, vitamins, coffee, snacks, contact lenses. The customer would buy these things anyway, the subscription removes friction and often provides a small discount. Retention is structurally favourable because the underlying need does not disappear, but it is bounded by alternatives: the customer can switch to a competitor, buy in bulk at a warehouse club, or use Amazon Subscribe and Save.

Curation subscriptions solve a discovery problem: Bespoke Post for men's accessories, Stitch Fix for clothing, BarkBox for new dog toys monthly, Birchbox-era beauty sampling. The customer values the curation novelty and the surprise factor. Retention is structurally unfavourable because the novelty exhausts: by month 6, the curator's range has been thoroughly sampled, and the customer increasingly perceives the box as duplicative.

The churn rate distribution reflects this structural split. Replenishment subscriptions cluster at 6 to 10 percent monthly churn for well-run operators, with Chewy autoship and Amazon Subscribe and Save sitting in the 4 to 6 percent range due to convenience-led economics. Curation subscriptions cluster at 10 to 18 percent monthly churn even for category leaders, with the cliff often hitting at month 4 to 6 when novelty fatigue sets in.

The strategic implication: curation subscriptions have to evolve into replenishment or community before the novelty cliff hits, or accept that the LTV ceiling is roughly 6 to 12 months per subscriber. Stitch Fix evolved toward style-personalisation (a quasi-replenishment proposition); Bespoke Post added a marketplace that lets subscribers buy individual items they have sampled; BarkBox added dental and supplement subscriptions that operate on replenishment economics.

Public reference points

Several large DTC subscription operators have disclosed enough through SEC filings, earnings calls, or third-party measurement to provide useful benchmarks:

• Stitch Fix: Approximately 65 to 75 percent annual retention historically. Public 10-Ks disclose active client counts and revenue per client. The numbers have compressed since 2022 as the curation novelty wave receded.
• HelloFresh: Approximately 50 to 60 percent annual retention. Disclosed in earnings via the active customer count. Meal-kit retention is notoriously weak because the meal-prep enthusiasm wanes faster than other replenishment categories.
• Chewy autoship: Approximately 85 to 90 percent annual retention. Disclosed indirectly through Chewy's autoship attach rate (now over 75 percent of revenue). Pet food is a structurally favourable replenishment category due to feeding consistency.
• BarkBox (BARK Inc): Approximately 60 to 70 percent annual retention. Disclosed in 10-Ks via active subscription counts. Pets-focused curation with strong community attach.
• Athletic Greens / AG1: Estimated 70 to 80 percent annual retention based on industry estimates (privately held, no public disclosure). Vitamin / supplement replenishment with strong brand-led discovery.
• Dollar Shave Club: Estimated 55 to 65 percent annual retention historically. Unilever divested in 2023 reportedly due to the difficulty of growing the subscription engine at acceptable economics.

The pattern: replenishment subscriptions with high consumption frequency (pet food, supplements) and weak alternatives achieve the best retention. Curation subscriptions with novelty as the primary value proposition achieve the worst retention. Hybrid models that evolve from curation toward replenishment usually outperform pure-curation peers.

Three-layer cost for subscription boxes

Layer one (direct MRR loss): Critical to think in gross-margin contribution terms rather than revenue terms because physical-goods subscriptions carry significant cost of goods. A $40 monthly box with 50 percent gross margin produces $20 of contribution per subscriber per month, so a churned subscriber loses $20 of monthly recurring contribution rather than $40.

Layer two (wasted CAC): DTC subscription CAC ranged $30 to $150 per subscriber in the 2020-22 period before iOS 14.5 attribution changes inflated paid social acquisition cost. Post-2022, the effective CAC is often higher because attribution is muddier. A churned subscriber who acquired at $80 CAC and stayed only 3 months at $20 contribution per month has produced $60 of contribution against $80 of CAC: a structural loss before any other consideration.

Layer three (forfeited expansion): Minimal in standard subscription boxes because expansion is rare. The exceptions are multi-product attach (Stitch Fix Men + Stitch Fix Women, BarkBox + Bark Bright dental, AG1 + travel packs) and family-plan-equivalent attach (multiple pets, multiple delivery addresses). For operators with strong cross-product motion, this layer can add 15 to 25 percent to the per-subscriber LTV calculation.

For a 100,000-subscriber subscription box at $40 monthly ACV and 50 percent gross margin: 8 percent monthly churn produces approximately $19M of annual three-layer cost when you include wasted CAC. Drop churn to 5 percent monthly and total cost falls to roughly $11M annually. The $8M annual delta typically justifies $2M to $4M of retention investment.

What works for subscription box retention

The retention interventions that consistently work for subscription boxes are model-specific:

For replenishment subscriptions: Delivery-frequency calibration is the single biggest lever. Give subscribers explicit control over the cadence and surface usage-based recommendations. A subscriber receiving boxes faster than they consume the product will churn within 60 days; a subscriber receiving boxes slower than they consume will disengage from the subscription mental model. Chewy autoship demonstrates the institutional version: subscribers can pause, skip, accelerate, or decelerate any individual product within their autoship, all without contacting support.

For curation subscriptions: Novelty replenishment through deep curator inventory and member-driven preferences. The cliff at month 4 to 6 hits when the curator's catalogue exhausts; operators that have deep enough catalogues and personalisation engines to keep finding novel items can extend retention by 6 to 12 months. Bespoke Post invests heavily in catalogue breadth precisely for this reason.

For both models: Annual subscription upsell with material discount. Converting 20 percent of monthly subscribers to annual at a 20 percent discount typically improves blended LTV by 30 to 50 percent because the annual cliff at month 12 is less damaging than the cumulative monthly attrition between months 1 and 12.

Cancellation flow optimisation. Subscription boxes typically see 15 to 25 percent save rates from well-designed cancellation flows: offer to skip the next delivery (preserves the subscription, addresses inventory pile-up), offer to swap product (addresses preference mismatch), offer to pause for 1 to 3 months (preserves the subscription through life-event disruptions), offer a discount (addresses price sensitivity). Operators that flow-test each branch see 30 to 50 percent save rates.

Frequently asked questions

Related reading on ChurnCost

• E-commerce churn calculator, the broader e-commerce retention picture.
• B2C subscription churn cost, the parent category.
• Mobile app subscription churn cost, the digital subscription counterpart.
• Gym subscription churn cost, the seasonality-loaded category.
• Involuntary churn, the payment-failure recovery framework.
• Retention playbook, the diagnostic framework.