The situation
A better-for-you snack brand sold through DTC subscriptions and retail. Subscription take rate supported a plan to increase acquisition spend.
The team reviewed subscription unit economics before increasing that spend.
The problem
The model didn't reflect actual subscription discounts, COGS, or repeat-order value.
One product line had discounts of 37% to 69%, while the model assumed 7%. Subscription COGS was 39.8% of net revenue, compared with 24.4% for the online store. The model used a 1.05× repeat-basket index, but the observed index was 0.73×. Subscription AOV was $34.30, compared with $46.75 for the online store.
Using the actual repeat basket, the model overstated repeat revenue by 43%. A separate error put the customer-acquisition ceiling $15 too low because trial samplers were blended into the subscriber average. The two errors distorted different decisions: how profitable subscription looked, and how much the business could pay for a genuine subscriber.
What we did
Rebuilt subscription economics with actual discounts, COGS, and repeat-order value.
Separated trial samplers from subscribers when setting the allowable acquisition cost.
Changed pricing and discounts for the product line with the deepest promotions.
Recalculated contribution margin by product to guide the product-mix decision.
The results
Subscription payback went from 12 months to 5.
Subscription average order value moved from $30.93 in May to $32.82 in June to $35.78 in July — a 15.7% increase across the period.
The team stopped the planned subscription spend increase after the corrected economics showed a loss at the current acquisition terms.