How should a brand measure attribution on subscription repeat orders?
Repeat subscription revenue did not come from this month's ads, but this month's ads did not create the subscriber either. Split the credit: acquisition gets the signup, retention gets the renewals.
Subscriptions break standard attribution
Standard attribution asks which touchpoint caused an order. For a subscription's fifteenth renewal, the honest answer is: the product, the habit, and the original signup experience, in some unknowable ratio. Crediting the renewal to this month's brand campaign is fiction. Crediting it to nothing leaves a huge share of revenue unattributed and makes retention work invisible. Subscription brands need a split model, not a single answer.
Separate acquisition from retention revenue
The foundational split is between the first order and everything after. The signup order is acquisition: credit it to the marketing that drove the trial, with whatever multi-touch model the brand uses. Every renewal after that is retention revenue. It belongs to a different budget, a different team, and a different measurement frame. Mixing them lets acquisition take credit for retention's work and hides whether the subscription program is actually healthy.
This split changes reported numbers dramatically. A brand where subscriptions are half of revenue will see its acquisition ROAS fall when renewals are removed from the acquisition column. That fall is the truth arriving. The acquisition team was never generating those renewals; they were harvesting them.
Measure the signup, then measure the cohort
Acquisition measurement for subscriptions should be cohort-based from the start. Do not just count signups per channel; track what each signup cohort is worth at 3, 6, and 12 months. A channel that drives cheap signups who cancel in month two is not a cheap channel. A channel with expensive signups who stay for two years is not expensive. Payback period and cohort lifetime value are the real acquisition metrics for subscriptions.
This requires joining marketing touch data to subscription billing data, which is where most brands stall. The subscription platform knows the renewals; the ad platform knows the signup touch. Until those are joined at the customer level, usually in the warehouse, attribution stays at the level of anecdote.
Attribute retention work to retention metrics
Renewals are not automatic; they are earned by the product, by lifecycle email, by the pause-and-skip experience, by win-back offers. Measure retention efforts against retention outcomes: churn rate, reactivation rate, expansion (upsell and cross-sell) revenue. An email flow that saves ten percent of at-risk subscribers has a measurable revenue impact, and it should be credited there, not smeared across acquisition channels.
Watch for the cancellation-flow trap. Brands that make cancellation hard keep short-term revenue and destroy the metrics that matter: they inflate subscriber counts with hostages, and the eventual chargebacks and complaints cost more than the saved renewals. Measure voluntary retention separately from total retention. The gap between them is the hostage count.
Handle the hybrid cases explicitly
Some orders sit between acquisition and retention: a subscriber's one-off add-on purchase, a gift subscription bought by an existing customer, a lapsed subscriber who returns through a paid ad. Write rules for these in advance. A common workable set: add-ons credit to retention (the relationship drove the purchase), reactivated subscribers credit to whatever reactivated them (often a win-back campaign, which is retention spend), and gift purchases credit to acquisition (a new customer relationship is forming).
The rules matter less than their consistency. Attribution arguments inside subscription brands usually trace back to two teams measuring the same revenue with different definitions. One shared definitions document ends most of those arguments.
Report the two numbers side by side
The executive report for a subscription brand shows acquisition efficiency (cohort payback, CAC against 12-month value) next to retention health (churn, net revenue retention, expansion). Neither number alone tells the story. A brand with great acquisition and terrible retention is filling a leaky bucket; a brand with great retention and weak acquisition has a product people love that nobody finds. The split model does not just measure better. It diagnoses.