Subscription Model or One-Time Purchase: How I Decide Per Product
I ran this decision at a DTC brand I scaled from $100K to $3M+. We had 14 SKUs. Six went subscription. Eight stayed one-time purchase. The split wasn't a vibe. It came from a checklist I still use today.
This post is that checklist. Subscription model vs one time purchase ecommerce isn't a company-wide decision. It's a per-product decision. Treat it like one.
What the problem actually looks like
Founders ask me "should we add subscriptions" like it's one switch. Then they flip it for everything in the catalog. Three months later:
- Subscriber churn hits 40% by month three on a product nobody actually reorders monthly
- Customer service gets flooded with "why was I charged again" tickets
- A $38 product with 11% margin is now sitting in a subscription flow built for a $90 product with 60% margin
- The one-time buyers who would've paid full price are getting trained to wait for the subscribe-and-save discount
The catalog looks optimized on paper. Revenue per product tells a different story.
Why it happens
Subscription logic gets applied as a company strategy instead of a product fit test. Someone reads that subscriptions increase LTV by 2-3x industry-wide, and that number becomes the justification for everything in the store.
But LTV lift only works if the product has a real, predictable consumption cycle. A supplement you take daily has one. A phone case does not. When you force subscription onto a low-frequency product, you're not building loyalty. You're building a billing dispute generator.
The other mechanism is margin blindness. Subscription programs cost money to run: discount given (usually 10-20%), payment processing on recurring failed cards, churn management, win-back email sequences. If the base margin can't absorb that cost and still be profitable, the subscription is subsidizing itself out of existence.
What we do about it
Here's the actual per-product checklist, in order. A product needs to clear most of these before it gets a subscription option.
- Consumption cycle is real and predictable. Can I tell you, within a two-week window, when this customer will run out? If yes, subscription candidate. If "it depends," no.
- Gross margin is 55% or higher. Below that, a 15% subscribe-and-save discount plus processing fees eats the profit. I don't subscribe anything under 50% margin, full stop.
- Repeat purchase rate without a subscription offer is already above 20%. If people aren't naturally coming back, a subscription won't manufacture demand. It just automates a purchase that wasn't going to happen.
- Unit price supports a discount. A $15 item can't absorb a 15% cut and still cover acquisition cost. A $60+ item usually can.
- Low return/exchange rate. Subscriptions on high-return products (apparel, sizing-dependent goods) create refund headaches on recurring billing. Keep those one-time.
- Customer doesn't need to choose a variant each time. If flavor, size, or color rotates based on mood, subscription friction goes up and conversion goes down. Build a "choose your box" flow or skip it.
If a product clears four or more of these six, I test subscription. If it clears two or fewer, I don't bother. The middle ground gets a "set a reminder" email flow instead, which captures some of the reorder lift without the billing complexity.
One more operational detail: we always soft-launch subscription on one SKU at a time, watch 90 days of churn data, then roll to the next candidate. Never launch subscription catalog-wide in one week. I've seen that create a support ticket spike that took a team of three two months to dig out of.
What it costs to ignore
On the infrastructure project I marketed, a $2.2B build, we had a tiered communications product (reports, alerts, briefings) that stakeholders could pay for or get packaged with their contract. The team wanted to subscribe everything by default to "lock in retention." I pushed back on two low-frequency reports because nobody consumed them more than twice a year. We kept those one-time purchase. Retention on the real subscription products stayed clean because we didn't dilute the program with mismatched items. Contrast that with the DTC brand: before I built the checklist, we'd subscribed a seasonal product with no repeat logic behind it. Churn hit 52% in 60 days. Refund requests cost us about $9,400 in processing fees and manual refunds that quarter, plus the support hours to handle it. That's the real cost of ignoring product-level fit: it's not lost opportunity, it's active cash leaving the business to clean up a mismatch you created.
The other cost is quieter. Subscription discounts train your best customers to never pay full price. If you add subscribe-and-save to a product people would've bought at full margin anyway, you just gave away 15% for nothing. That's a margin decision, not a growth decision, and it rarely gets modeled before launch.
The most common mistake
The single biggest mistake: deciding subscription model vs one time purchase ecommerce at the brand level instead of the SKU level. "We're a subscription brand" or "we're not a subscription brand" is the wrong frame. Some of your products want it. Some actively suffer from it. Decide product by product, every time, using margin and consumption cycle as your filters, not category trends or what a competitor is doing.
FAQ
Should every new product launch with both options?
No. Launch one-time purchase only. Get 90 days of repeat purchase data. Then decide if subscription fits using the checklist above. Adding subscription to an unproven product just adds billing complexity to a product you don't understand yet.
What discount should I offer for subscription?
10-15% is the range that works for most consumable products without destroying margin. Anything above 20% usually means you're compensating for weak product-market fit with price, which doesn't fix retention, it just delays the churn conversation.
Practical takeaway: Pull your top 10 SKUs by revenue. Run each one through the six-point checklist above. You'll probably find two or three that should be subscription and aren't, and one or two that are subscription and shouldn't be. Fix those first before you touch anything else in the catalog.