Marketing ROI With vs Without Subscription Revenue Counted
For subscription-heavy DTC brands, counting only first-order revenue understates ROAS — but counting full LTV overstates it. Here's how to pick the right window, with worked examples for skincare, supplements, and coffee.
Quick answer
For subscription DTC brands, report marketing ROI two ways: first-order ROAS (cash recovered on day 1) and 90-day contribution ROAS (cash recovered through the first reorder or refill). Use first-order to govern paid-media cash flow; use the 90-day number to decide whether to scale. Never report full LTV ROAS to finance — it hides payback risk.
Marketing ROI With vs Without Subscription Revenue Counted
The choice between attributing only first-order revenue, a fixed-window contribution, or full LTV to a marketing campaign when computing ROI.
When a meaningful share of revenue comes from subscriptions or repeat purchases, first-order ROAS undercounts what a campaign is worth — and full LTV ROAS overcounts it relative to the cash you actually have. The practical question for any subscription-heavy brand is which revenue window to credit back to acquisition spend.
Three conventions dominate: first-order revenue only (most conservative, easiest to audit), a fixed contribution window such as 90 or 365 days (a balance between predictability and recognising repeat value), and full modelled LTV (most generous, most prone to forecast error). The right choice depends on payback tolerance, cohort stability, and what decision the number is informing.
The tension is simple. A skincare brand selling a €40 serum on a 90-day refill cadence has a very different revenue curve from a coffee brand on a weekly auto-ship. Crediting only the first order treats both campaigns the same. Crediting full LTV treats them as if every acquired customer stays forever.
The decision matters most when ad CPMs rise. Channels that look unprofitable on first-order ROAS often clear comfortably once the second and third orders are counted — but only if your retention curve actually delivers them.
Why first-order ROAS misleads subscription brands
First-order ROAS treats each acquisition like a one-shot transaction. For a supplements brand where 55% of new customers subscribe and the average subscriber stays 7 months, that view writes off most of the campaign's economic contribution.
Concrete case. A Meta prospecting campaign spends €10,000 and acquires 250 first orders at €38 AOV. First-order revenue is €9,500 — a ROAS of 0.95, which looks like a loss. Counted across 90 days with one expected refill at 40% retention, contribution revenue rises to roughly €13,300, a 1.33 ROAS. The campaign is profitable; the first-order view just couldn't see it.
Two numbers, not one
Don't replace first-order ROAS with windowed ROAS — report both. First-order tells you whether you can afford the campaign this month. Windowed ROAS tells you whether it's worth running at all. Collapsing them into a single number is how brands end up cash-constrained mid-quarter.
Picking the right contribution window
The window should match your typical reorder cadence and your payback tolerance. A weekly coffee subscription can credibly use a 60-day window — you'll see four to eight orders by then. A quarterly skincare refill needs at least 120 days to capture the second order with any confidence.
Pair the window with a retention assumption drawn from real cohort LTV curves, not a flat average. If 30-day repeat is 28% for your category and your cohort sits at 22%, use 22%. The cleanest source is your own historical orders segmented by acquisition month.
A defensible default for most subscription DTC brands: report first-order ROAS and a 90-day contribution ROAS that counts realised revenue only (not modelled forward LTV). This keeps the number auditable and stops finance arguments about which forecast to trust.
Benchmark: how the window changes the verdict
Same campaign, three windows: typical ROAS by subscription category
| Category | First-order ROAS | 90-day ROAS | 365-day ROAS | Avg orders / yr |
|---|---|---|---|---|
| Skincare (quarterly refill) | 0.9 | 1.4 | 2.6 | 2.8 |
| Supplements (monthly) | 0.8 | 1.9 | 4.2 | 6.5 |
| Coffee (bi-weekly) | 0.7 | 2.3 | 5.1 | 11.0 |
| Apparel (non-subscription) | 1.2 | 1.5 | 1.8 | 1.6 |
Notice the pattern. Higher reorder cadence means a wider gap between first-order and annual ROAS — and a stronger case for using a contribution window when judging channel performance. A coffee brand looking only at first-order ROAS would shut down channels that are actually their best long-term acquisition sources.
Worked examples by category
Skincare, €52 AOV, 90-day refill. €15,000 Meta spend, 300 first orders. First-order revenue €15,600 (1.04 ROAS). 45% return for a second order at 90 days, adding ~€7,020. 90-day contribution ROAS: 1.51. The channel is healthier than first-order suggests but isn't a runaway — scale cautiously.
Supplements, €34 AOV, 60% subscribe at checkout. €20,000 Google spend, 580 first orders. First-order revenue €19,720 (0.99 ROAS). Of 348 subscribers, ~80% are still active at day 90 with an extra two refills each: ~€18,950 added. 90-day ROAS: 1.93. Coffee, €28 AOV, bi-weekly: €8,000 spend brings 290 first orders, ~€16,200 in 90 days — a 2.03 ROAS.
When full LTV ROAS is appropriate (and when it isn't)
Full LTV ROAS belongs in annual planning, board decks, and channel-mix strategy — places where you're modelling a portfolio over 12-24 months and have stable cohort LTV curves to back the forecast. It does not belong in weekly paid-media reviews. The forecast error compounds and the cash impact is invisible.
A useful rule: if the decision is 'how much should I spend tomorrow,' use first-order ROAS. If the decision is 'should this channel be in our mix at all,' use a 90- to 365-day contribution ROAS. If the decision is 'what's our acquisition budget for next year,' full LTV ROAS is fair — bounded by a sensitivity analysis on the retention assumption.
Frequently asked questions
Yes, but as a separate metric, not as a replacement for first-order ROAS. Report first-order ROAS for cash-flow decisions and a windowed contribution ROAS (typically 90 or 365 days) for channel-mix decisions. Full modelled LTV ROAS is only safe when your cohort LTV curves are stable across at least 6-12 months of data.
Take campaign spend in the period, then attribute realised revenue from acquired customers across your chosen window (90 days is the common default). Divide attributed revenue by spend for ROAS, or subtract COGS and fulfilment first for true contribution ROI. The key is using realised orders, not forecast LTV.
Match the window to your reorder cadence. Weekly or bi-weekly subscriptions (coffee, pet food): 60-90 days. Monthly (supplements, household): 90-180 days. Quarterly (skincare refills, vitamins): 180-365 days. Shorter windows are more conservative; longer windows capture more revenue but introduce more retention-forecast risk.
The platform-reported ROAS will always be near-real-time first-order. Your contribution ROAS is a finance-side calculation that joins ad-platform spend with your order database via customer ID or email. Most subscription brands maintain it in a BI tool or spreadsheet, refreshed weekly.
Count only realised revenue — orders that actually shipped and weren't refunded. If a customer cancelled after two refills, those two refills count; the projected third does not. This is what makes a windowed contribution ROAS more defensible than full LTV ROAS for operational decisions.
For directional channel comparisons, gross revenue ROAS is fine. For real profitability decisions — especially when scaling spend — use contribution margin (revenue minus COGS, payment processing, fulfilment, and returns). Subscription brands often discover that a 2.0 gross ROAS is only a 1.1 contribution ROAS once refills and returns are netted out.
Attribute all repeat revenue to the original acquisition channel for first-90-day contribution ROAS — the question is whether acquisition spend was worth it. Separate that from retention-channel performance (email, SMS, loyalty), which should be measured on its own incremental lift, not by re-attributing already-acquired customers.
Payback period is the inverse question: rather than asking 'what ROAS at 90 days,' it asks 'how many days until ROAS hits 1.0.' Both metrics use the same windowed revenue logic. Subscription brands typically target a 4-6 month payback; coffee or pet-food brands can sometimes hit 2-3 months.
Only if you've validated the model against at least two full years of cohort outcomes and you report a confidence interval, not a point estimate. Most teams that try this end up overstating ROAS by 20-40%. Realised-revenue windowed ROAS is less precise but far more honest.
Quarterly at minimum, and immediately after any pricing, packaging, or onboarding change. Retention curves drift, and a stale assumption can keep a channel looking profitable for months after its real economics have turned. Re-fit the curve from your cohort LTV data each quarter and document the change.
Track CAC, channels, and funnel conversion in one place
Metricuno connects ad spend, funnel events, and revenue so you can see CAC by channel, cohort, and campaign — without stitching together five tools.