Modeling Subscription vs One-Off Retention Lift in the Same Catalogue
Mixed catalogues break single-rate retention models. Here's how to weight subscription and one-off retention separately and report a defensible blended LTV.
Quick answer
Model subscription SKUs and one-off SKUs as two separate cohorts with their own retention curves, then blend the resulting LTVs by revenue share — not by customer count. A typical refill-plus-apparel store sees subscription retention of 55–75% at month 12 and one-off repeat rates of 20–35%; blending them at the customer level hides both.
Subscription vs One-Off Retention Modeling
Splitting retention inputs by purchase mode (subscription vs one-off) and combining them into a single blended LTV for a mixed catalogue.
When a catalogue mixes consumable refills (subscription) with one-off goods (apparel, accessories, gifts), the two purchase modes have fundamentally different retention curves. Subscription customers churn on a contractual schedule; one-off buyers churn through silence. Feeding a single average retention rate into the Retention Lift LTV Calculator under-states subscription LTV and over-states one-off LTV at the same time.
Split-modeling solves this by computing two retention rates and two LTVs, then weighting them by revenue contribution to produce a blended LTV that operators can defend to finance and use in CAC payback decisions.
The mistake most operators make is averaging retention at the customer level. If 30% of buyers are on subscription with 70% annual retention and 70% are one-off with 25% repeat, the weighted-customer average comes out near 38% — a number that flatters neither cohort and predicts a future that never arrives.
Revenue weighting is the correct lens because LTV is a revenue forecast, not a headcount forecast. A subscription customer typically generates 3–6× the annual revenue of a one-off buyer, so their retention behaviour dominates the blended outcome even when they're a minority of the file.
Why the two modes need separate inputs
Subscription retention is gated by the cancel button. The decay is steep in months 1–3 (post-trial churn, billing failures, sample-size remorse) and then flattens — a beauty refill program might lose 40% of month-1 subscribers but retain 80% of whoever survives to month 4.
One-off retention is gated by intent. A customer who bought a swimsuit in June has no contractual relationship — they re-appear when the next need does. The curve is flatter, longer, and far more sensitive to lifecycle marketing than to product satisfaction.
Don't use the same churn formula
Subscription churn is monthly and compounding; one-off "churn" is an inferred lapse after N months of silence (usually 2× the median inter-purchase interval). Treating them with one formula will systematically misprice your CAC ceiling.
How to enter split inputs in the calculator
In the Retention Lift LTV Calculator, run two passes. First pass: subscription cohort only — use monthly retention, the subscription AOV, and the actual gross margin on refills (usually 60–75% on consumables).
Second pass: one-off cohort — use annual repeat rate as the retention input, the one-off AOV (typically higher per order but lower frequency), and the apparel/accessory margin (45–60% after returns).
Then compute revenue share: subscription_revenue / (subscription_revenue + one_off_revenue) over a trailing 12-month window. Multiply each LTV by its share and sum. That's your blended LTV — the number to compare against blended CAC.
Typical retention curves by catalogue mix
Retention and LTV benchmarks by purchase mode and category
| Category | M12 retention (sub) | 12-mo repeat (one-off) | AOV (sub) | AOV (one-off) | Indicative LTV |
|---|---|---|---|---|---|
| Beauty refills + makeup | 62% | 28% | €34 | €52 | €280 / €95 |
| Pet food + accessories | 74% | 31% | €48 | €38 | €420 / €72 |
| Vitamins + wellness extras | 58% | 22% | €42 | €61 | €310 / €88 |
| Coffee + brewing gear | 68% | 26% | €28 | €85 | €240 / €130 |
| Apparel basics + capsule drops | 55% | 35% | €55 | €95 | €295 / €185 |
Read the LTV column as subscription / one-off, not as a blend. Notice that pet and beauty have the widest gap — those are the categories where averaging the two retention rates does the most damage to your forecasting.
Segmenting the file for clean inputs
Tag every customer by their first-order mode. "Subscription-first" customers behave differently from "one-off-first who later subscribed" — the latter group has 15–25% higher subscription retention because they've already validated the brand on a low-commitment order.
If your historical GA4 import lands on day one, you can build these cohorts retroactively rather than waiting six months for fresh data. Match subscription_started events against prior purchase events and you have a clean lineage column for the calculator.
Reporting the blended LTV to operators
Always report three numbers, not one: subscription LTV, one-off LTV, and the revenue-weighted blend. Finance teams will challenge a single blended figure; showing the components makes the assumption stack auditable and the CAC ceiling defensible per channel.
When you brief paid media, attach the LTV that matches the landing page. A subscription-quiz ad should be evaluated against subscription LTV, not the blend. A capsule-drop campaign uses one-off LTV. The blend is for board-level unit economics, not channel bidding.
Common questions about split retention modeling
Always by revenue contribution over a trailing 12 months. Customer-count weighting under-weights subscription customers, who typically generate 3–6× the revenue of one-off buyers despite being a smaller share of the file.
Split the order at the line-item level for revenue attribution, but classify the customer by their first subscription event (or by current active subscription status). Their retention curve follows the subscription cohort, not the mixed-basket cohort.
Treat paused subscribers as retained if the pause is ≤90 days and reactivation rate is above 60%. Beyond that window, count them as churned but track reactivation separately — it usually behaves more like a win-back than continued retention.
Use the 12-month repeat purchase rate from your own data — typically 20–35% for apparel and accessories, higher for replenishable categories like skincare. If you don't have 12 months of history, the Retention Lift LTV Calculator will accept a 6-month input scaled with a decay assumption.
Yes — use the post-discount AOV and the post-discount gross margin. A 20% subscriber discount on a 65%-margin product drops effective margin to 56%, which compounds across the full retention curve and changes the LTV by 12–18%.
Quarterly for the inputs, monthly for the outputs. Retention curves shift slowly but AOV and margin move with promotional calendars, so the blended LTV needs a monthly refresh to stay aligned with current CAC bids.
Yes if subscription LTV is more than 2× one-off LTV, because that small revenue share punches above its weight in retention forecasting. Below 5% of revenue and below 2× LTV multiplier, a single blended input is acceptable.
Net out returns from the AOV and gross margin inputs before computing LTV. Don't adjust the retention rate itself — a returner who buys again is still a retained customer, just one with a lower contribution margin per cycle.
The framework transfers, but the retention curve shapes are even more divergent — wholesale has contracted reorder cycles, DTC has intent-driven returns. Treat them as fully separate businesses with separate CAC and LTV, not as a blended mix.
Use the blend for portfolio-level CAC payback (the board-level number), and the unblended component LTVs for channel-level payback. A subscription-acquisition channel should be judged on subscription payback in months, not on the blended payback that one-off buyers drag out.
Start tracking the metric that matters
Free for 30 days. No credit card. Connect your site in 4 minutes and see the one metric driving revenue.