A subscription box business reports two numbers to its investors — monthly recurring revenue and active subscribers — and both can rise for a year while the business gets worse. A first box discounted to 50 percent acquires subscribers who leave at month two; a blended churn rate averages the loyal base with the churning cohort and reports 6 percent when the new cohorts are losing 25 percent a month; failed card payments are counted as cancellations and never retried; and annual prepaid plans, paid in December, are recognized as revenue in December while twelve boxes ship through November. The economics live in the cohort tables and the deferred-revenue schedule, and the business that has built them can tell growth from replacement.
Why is blended churn the wrong number?
Blended churn averages cohorts at different ages. A business with 20,000 subscribers, half acquired more than a year ago at 3 percent monthly churn and half acquired in the last three months at 18 percent, reports a blend near 10 percent and learns nothing from it; the loyal base is fine and the recent acquisition is failing, and the blend says neither.
Cohorts by acquisition month and by offer. Subscribers who joined in the same month, tracked monthly for the share still active; by the offer they joined on — full price, 25 percent off the first box, 50 percent off, free first box — and by channel; the curves show the month-two cliff on discounted cohorts and the gentle slope on full-price ones, and the business learns what its acquisition is actually buying.
Retention at month three predicts the rest. The share of a cohort active at month three correlates with its lifetime far better than the first-month number; a cohort at 55 percent at month three will usually retain its remaining subscribers for a long tail; one at 30 percent will not; the month-three rate by offer is the metric to watch weekly.
How does the discounted first box buy churn?
The first-box discount is acquisition cost, not revenue. A US$40 box offered at US$20 is US$20 of acquisition cost on top of the advertising that drove the signup; the business's acquisition cost per subscriber includes it; and a cohort acquired at 50 percent off that churns at month two has produced one discounted box and one full-price box against the full acquisition cost — a loss on nearly every subscriber.
Offer economics by cohort. For each offer: acquisition cost including the discount, retention curve, boxes per subscriber over twelve months, contribution per box, and lifetime contribution; the deepest discount usually acquires the most subscribers and the least value; the offer the business should scale is the one whose cohort pays back, which is often the shallowest.
What is involuntary churn, and what does dunning recover?
A failed payment is not a cancellation. Expired cards, declined transactions, insufficient funds, and bank changes; in subscription commerce they run 20 to 35 percent of total churn; most businesses cancel the subscriber after one or two failures and count the loss as churn.
The dunning process. Pre-expiry card updates, automated retries on a schedule tuned to issuer behavior, a card-update prompt by email and in the account, a pause rather than a cancellation after the retry window, and account-updater services that refresh card details with the networks; businesses that run the full process recover 40 to 60 percent of involuntary churn, which is frequently the single largest retention lever available and the cheapest.
The metric. Involuntary churn as a share of total; recovery rate by step; revenue recovered monthly.
How are prepaid plans accounted for?
An annual plan is a liability until the boxes ship. US$420 collected in December for twelve monthly boxes is US$420 of deferred revenue, recognized at US$35 per box as each ships; a business recognizing it in December overstates December and understates the following eleven months, and its monthly recurring revenue, margin, and seasonal pattern are all wrong.
The deferred-revenue schedule by plan. Balances by plan type and by month of expected shipment; the cost of fulfilling the deferred boxes as a disbursement ahead; the cash from prepaid plans read against the obligation — because the December cash is strong and the following year's shipments are already paid for; breakage on prepaid boxes never claimed, recognized on a policy.
Quarterly and semi-annual plans follow the same rule, and so do gift subscriptions, which are a liability until redeemed and shipped.
What does a box actually cost?
Cost per box, fully loaded. Product cost — the items in the box at acquisition cost, including the samples and inserts vendors supply at a price; packaging and the custom box; fulfillment labor or third-party fulfillment fees; outbound shipping at the actual carrier cost including surcharges and dimensional weight — the line most businesses under-count because the subscriber pays "free shipping"; payment processing; and the replacement cost of damaged or lost boxes.
Contribution per box by plan. Price per box by plan (monthly, quarterly, annual) less cost per box; annual plans priced at a discount carry lower contribution per box and must be justified by retention and cash timing; the mix of plans shapes blended contribution.
Product cost variability. Boxes curated monthly have a product cost that varies by month; the business sets a product cost budget per box and tracks the curation against it, because a box that over-delivers by US$6 of product across 20,000 subscribers is US$120,000 of margin in one month.
What is lifetime contribution, and how should it set the acquisition budget?
Lifetime contribution per subscriber by cohort. Boxes per subscriber over the cohort's observed life times contribution per box, less the acquisition cost including the first-box discount; by offer and channel; a cohort worth US$84 acquired at US$61 pays back; one worth US$38 acquired at US$61 does not, regardless of how many subscribers it added.
Payback in boxes. Acquisition cost divided by contribution per box gives the boxes to break even; against boxes per subscriber by cohort; a cohort that needs 4.2 boxes to pay back and averages 3.1 is a loss on every subscriber it added.
The acquisition budget is the output. A target payback and the cohort curves decide how much to spend in each channel on each offer; the business stops buying subscribers from cohorts that do not pay back, and the growth rate becomes whatever the paying cohorts support.
How is growth separated from replacement?
Gross additions, churn, and net additions by month. A business adding 4,000 subscribers and losing 3,600 is growing by 400 and replacing 90 percent of its additions; the replacement rate — churn over gross additions — tells the owners how much of the acquisition spend is treading water; a business replacing its base every eleven months is not growing, it is running.
The monthly subscriber report. Active subscribers by cohort age; gross additions by offer and channel; churn split voluntary and involuntary; dunning recovery; month-three retention by offer; contribution per box by plan; deferred-revenue balance and the fulfillment cost ahead; lifetime contribution and payback by cohort; replacement rate.
Why is this CFO work?
Cohort tables and offer economics require subscription, billing, and marketing data joined by subscriber; the dunning process is a revenue-recovery control; deferred revenue by plan is a balance-sheet discipline; cost per box with real shipping is a unit-cost build; and the acquisition budget set by payback is capital allocation the founders need imposed on the growth plan. The product and marketing teams own the box and the offers; the CFO owns the economics of the subscriber.
Worked example
A subscription box business, 22,000 active subscribers, US$10.4 million revenue, growing on a 50 percent-off first-box offer through paid social.
| Metric | Found | After twelve months |
|---|---|---|
| Blended monthly churn | 9.4% | 5.8% |
| Month-three retention — 50%-off cohorts vs. full-price cohorts | 31% vs. 58% | 50% offer withdrawn; 20% offer at 47% |
| Involuntary share of churn; recovery | 29%; none (cancelled after one failure) | 24%; 52% recovered |
| Annual prepaid plans | Recognized at sale; US$1.1M in December | Deferred; US$1.1M liability with fulfillment cost scheduled |
| Cost per box — reported vs. fully loaded (shipping at actual) | US$21.40 vs. US$27.90 | US$26.10 (packaging and carrier renegotiated) |
| Contribution per box (US$39 monthly plan) | US$11.10 actual vs. US$17.60 believed | US$12.90 |
| Lifetime contribution — 50%-off cohort vs. full-price | US$29 vs. US$91 | — |
| Acquisition cost including discount — paid social 50%-off | US$68 | US$54 on 20% offer |
| Payback in boxes — paid social 50%-off | 6.1 boxes vs. 2.9 delivered | 4.2 vs. 5.4 delivered |
| Replacement rate (churn / gross additions) | 88% | 61% |
Illustrative figures for a hypothetical organization; not a client's data.
The business had added 4,100 subscribers a month and lost 3,600, on an offer whose cohorts churned at month two and never paid back; it had booked US$1.1 million of December prepaid revenue that eleven months of boxes had yet to earn; and its cost per box was US$6.50 higher than the owners believed because shipping was carried as a marketing cost. The plan: the 50 percent offer withdrawn, a 20 percent offer tested with month-three retention as the gate; a full dunning process; deferred revenue by plan with the fulfillment cost scheduled; shipping and packaging renegotiated; a product cost budget per box; and the acquisition budget set by payback at four boxes. Subscriber growth slowed from 500 a month to 350; contribution after marketing moved from −US$210,000 to +US$940,000.