Subscriptions that survive month three.

Does your product deserve a subscription

The uncomfortable first question. Subscriptions work when the customer would have rebought anyway and you are removing the friction of remembering. They fail when the subscription exists to smooth your revenue rather than to serve a real repurchase rhythm.

• Replenishment is the strong case. Supplements, coffee, pet food, skincare, razors, anything consumed on a predictable clock. The customer has a genuine reorder date and you are meeting it. Replenishment programmes are consistently the lowest-churn category in the subscription data.
• Curation is the hard case. A surprise box every month is entertainment, and entertainment gets cancelled when budgets tighten. It can work, but it needs a much stronger content and community reason to stay, and it churns faster.
• Access and membership is a different product. Free shipping, member pricing, early drops. Often a better fit than a product subscription for brands whose catalogue is not consumable, and it does not carry fulfilment cost per cycle.

If you cannot state, in one sentence, why a customer would be worse off cancelling, you have a discount programme rather than a subscription, and it will churn like one.

The math nobody runs before launching

Two numbers decide whether the programme is a business.

Average lifetime is approximately one divided by monthly churn. At 10 percent monthly churn, the average subscriber lasts about ten cycles. At 7 percent, about fourteen. That is a 40 percent increase in lifetime value from a three point move, and it costs far less than growing acquisition by 40 percent. This is why churn work outranks funnel work once a programme is live.

Contribution per cycle has to survive the subscriber discount. The standard 10 to 20 percent subscribe-and-save discount applies to every future order, not just the first. If your margin only works because of full-price orders, a successful subscription programme can grow revenue while shrinking profit. Model it per cycle, including fulfilment and payment costs, before you set the discount.

Benchmarks for orientation rather than targets: industry reporting puts typical direct-to-consumer subscription churn in the high single digits per month, with the strongest programmes running roughly half that. Month one churn runs several times higher than steady-state churn in almost every programme, which is why the first cycle deserves disproportionate attention.

The month one cliff

The biggest single loss in most programmes happens before the second charge, and the causes are boringly consistent.

• The cadence is wrong. You set 30 days because it is tidy. The customer uses the product in 45. The second box arrives while the first is half full, and the rational response is to cancel. This is the most common cause and the easiest to fix: set the default interval from actual consumption, and let people change it in one click.
• The customer did not know they subscribed. A subscribe-and-save option preselected on the product page produces signups and then chargebacks. Anything that surprises someone at the second charge is worse than no signup at all.
• The first charge is the discount and the second is not. A deep introductory offer buys a subscriber who values the offer, not the product. Milder first-order discounts with a persistent subscriber benefit retain far better.
• Nothing happened between order one and order two. No confirmation of the next charge date, no reminder, no way to see the schedule without emailing support. Silence between cycles is where intent decays.

The third of churn that has nothing to do with satisfaction

Involuntary churn is a payment failing on a customer who fully intended to keep buying: an expired card, a reissued number after fraud, a bank declining a recurring charge, an insufficient balance on the day. Across subscription industry reporting it accounts for roughly a third of total churn, and it is the highest-return work available to a subscription programme because these customers are not trying to leave.

Four mechanisms, in rough order of value.

• Card account updater. The networks push updated card details when a card is reissued or expires, so the charge succeeds without the customer doing anything. It is the closest thing to free recovery, because it prevents the failure rather than reacting to it. Confirm your payment gateway and subscription app actually support and enable it; many programmes assume it is on and it is not.
• Intelligent retries. Not the same charge at the same hour for three days. Retry on a schedule that respects why cards decline: after payday, at a different time of day, spaced across the week. The retry strategy is a setting most merchants never open.
• A dunning sequence a human would respond to. Tell the customer their order is at risk, say what will happen and when, and give them a one-tap link to a hosted page to update the card. Not a login. Not a support ticket. Vague "there was a problem" emails do not get acted on.
• Pre-dunning. Email before the charge when a card is due to expire before the next cycle. Preventing the failure is worth more than recovering it.

Before you build any of this, measure it. Split your cancellations into voluntary and involuntary. Most brands have never looked, and are surprised by how much of their churn is a payments problem being reported as a retention problem.

Voluntary churn: give them everything except cancel

When someone genuinely wants to stop, the goal is not to trap them. It is to find out whether the thing they want is actually to stop, or to stop this particular charge on this particular date.

• Skip. The single highest-value option. Someone with too much stock does not want to cancel, they want this month off. Make it one tap and offer it before cancel.
• Reschedule and change frequency. Same logic. Most cancellations for "too much product" are cadence complaints.
• Swap. Boredom is a real cancellation reason for flavours, scents and variants. A swap keeps the subscription alive at full value.
• Pause with a date. Better than cancel, because it keeps the relationship and the payment method.
• Ask why, in one question. A single-select reason on the cancel flow, stored and reported. Without it you are guessing at which of the above to build.

The line to hold is that the cancel path must remain honest and easy to find. Burying it produces chargebacks, complaints and, increasingly, regulatory attention. Offer alternatives before cancellation, then let people cancel.

Native Shopify Subscriptions or a specialist app

Shopify's own subscriptions app is built directly on the platform's subscription APIs and runs through native checkout. Specialist apps such as Recharge maintain their own billing and customer data alongside your store, syncing through webhooks and APIs, and offer considerably more depth around portals, bundles, dunning configuration and analytics.

A reasonable rule: start native if you are launching one or two simple recurring products and want to find out whether the programme works at all. The native app handles that cleanly at no extra subscription cost. Move to a specialist when you hit a specific wall, and be able to name the wall. Common ones are dunning and retry configuration, subscriber portal depth, build-a-box and bundle logic, prepaid and gift terms, and cohort analytics you cannot get natively.

Two things to know before you choose. Migrating between subscription platforms later means moving stored payment credentials between vaults, which is a real project with a real timeline rather than a data export, so the switching cost is higher than for most app categories. And whichever you pick, the retention mechanics above matter more than the vendor. A well-run native programme beats a badly configured specialist one.

What to measure

• Cohort retention curves, not blended churn. Blended churn mixes a five-year subscriber with a first-cycle one and hides the month-one cliff completely. Group subscribers by signup month and watch each cohort decay.
• Voluntary against involuntary churn. The split tells you whether to work on the product or the payments stack.
• Cycle one to cycle two conversion. The single most diagnostic number in the programme.
• Payment recovery rate. What share of failed charges eventually succeed. If you do not know it, dunning is not configured.
• Skip and pause rates. Rising skips are an early warning that cadence is wrong, and they show up before the cancellations do.
• Contribution per subscriber lifetime. After discount, fulfilment, payment fees and support, against acquisition cost. The only number that says whether the programme is worth running.

Common mistakes

• Buying signups while the payments leak runs. Spending on acquisition before fixing failed-payment recovery is filling a bucket with a hole in it.
• One cadence for every customer. Consumption varies by person. A fixed 30 days manufactures cancellations.
• Preselecting subscribe on the product page. It flatters the signup number and produces chargebacks.
• Hiding cancellation. It converts a cancellation into a chargeback and a complaint, and it is a compliance risk.
• Discounting to a margin that does not survive success. Model the discount across the whole lifetime, not the first order.
• Watching blended churn. It is the metric most likely to tell you everything is fine while the newest cohort collapses.

FAQ