Two equipment repair companies can close a year with identical sales and be worth very different amounts. The first sells repairs. Every January it starts at zero and sells its way back to the same number. The second sells annual service plans with scheduled maintenance visits, and every January it opens with most of the year already committed. Same work, same customers, same market. The difference is who carries the risk of the next buying decision.
That transfer of risk is the real prize, and it is why so many owners want to design a business model around recurring revenue. But the mechanics people usually focus on, billing cadence and subscription software, are the least interesting part. The decisions that determine whether this works are made earlier, in the structure of the offer itself.
Three different things can recur, and only one of them is easy
Recurring billing recurs. Recurring value recurs. Recurring work recurs. These are separate, and almost every failed subscription business has confused at least two of them.
Billing is trivial to set up. Any payment processor will charge a card monthly. That is the part competitors’ articles spend half their word count on, and it deserves about a paragraph.
Value is the hard one. The customer has to receive something in month seven that is worth paying for in month seven. Not a memory of the onboarding, not a feeling of loyalty, something that actually arrives. When value does not recur, the model does not fail quietly. It fails with cancellations, chargebacks and complaints. The Federal Trade Commission reported that negative option and recurring-subscription complaints averaged nearly 70 per day in 2024, up from 42 per day in 2021. Those complaints came from somewhere.
Work is the one people forget. If delivering month twelve costs you exactly as much as delivering month one, you have bought predictability but not profitability. This is the quiet correction to a claim repeated everywhere: recurring revenue does not automatically improve margins. Shopify’s own documentation makes the point about replenishment subscriptions, which convert and retain well but carry thin margins. A monthly retainer with no leverage in it is simply a job with better scheduling.
So the test I would apply before anything else has three parts:
- Does the customer’s need come back on its own, without you having to manufacture a reason?
- Does it come back on a cadence you can predict within a few weeks?
- Does serving the same customer get cheaper, or at least no more expensive, over time?
Two out of three is workable. One out of three means you are about to bolt recurring billing onto a one-time problem, and the market will correct you within two renewal cycles. The gap between recurring value and recurring work is where the margin lives, and it is the same gap that determines whether your business model can scale at all.
Choose the archetype before you choose the price
“Subscription” is not a model. It is a billing arrangement shared by several models that behave nothing alike. Pick the wrong one and you will spend years fighting problems that belong to a structure you never needed.
Replenishment
The customer consumes something and needs more of it. Coffee, filters, contact lenses, cleaning supplies, lab consumables. Shopify’s framework notes that replenishment converts and retains better than other subscription types but operates on thinner margins. That tradeoff is the whole story. The renewal reason is convenience, and convenience is a weak moat against a competitor who undercuts you by fifteen percent. Replenishment works when you own the relationship or the logistics, not when you own the product.
Curation
The customer pays to be surprised. Boxes, selections, picks. Shopify flags this one as having high profit potential alongside high churn and real operational complexity. The margin is attractive because discovery is worth paying for. The churn is brutal because novelty has a half-life. If you choose curation, assume you are running an acquisition business with a subscription attached, and budget accordingly.
Access and membership
The customer pays for standing entry: member pricing, a community, a library, priority. Shopify describes this as offering greater perceived value and better bundling opportunities, but demanding more time investment. That time investment is the catch. Access models look passive and are not. Someone has to keep the room worth being in. The renewal reason here is identity and belonging, which is the most durable reason on this list and the hardest to fake.
Service retainers and maintenance plans
Underrated, and almost absent from the ecommerce-flavoured advice that dominates this topic. An accounting practice moving from annual filings to monthly advisory. A commercial landscaper selling a year-round grounds contract instead of seasonal jobs. An IT firm selling managed services instead of break-fix. The renewal reason is risk reduction: the customer is buying the absence of a problem. These models have the best margin trajectory of any archetype because expertise compounds while delivery time falls. They have the worst concentration risk, because losing three accounts out of forty can remove a quarter of your recurring base in one month.
Usage and metered
The customer pays for what they consume. Price tracks value and tracks your own costs, which is why it suits anything infrastructure-heavy. The weakness is that it hands your forecast to the customer. Buyers with procurement processes dislike unpredictable bills, and finance teams dislike them more.
Before moving on, write one sentence naming why a customer renews in month eight. Convenience, discovery, belonging, risk reduction, or consumption. If the sentence is hard to write, the model is not designed yet, it is just priced.
Price it as a combination, not a religion
Conventional advice says pick a dominant pricing model and commit. The recent evidence points the other way. Chargebee’s 2025 monetization study found that companies combining recurring subscriptions with usage, outcome-based or flat-fee elements were more than twice as likely to increase profit margins as those using purely usage-based pricing. In the same research, companies that added AI to their products and also rethought pricing were nearly twice as likely to achieve high growth as those that added AI without touching pricing.
Read practically, that suggests a base plus a variable. A committed monthly amount that covers your fixed delivery cost and gives both sides predictability, plus a metered component that captures the heavy users without punishing the light ones. The base protects your forecast. The meter protects your margin when a customer’s consumption triples.
Price movement is also now normal rather than exceptional. Chargebee reported that nearly 73% of subscription businesses planned price increases in 2024, up from 62% the year before. If your plan has no stated mechanism for adjusting price, you are designing a model that silently loses ground every year.
The annual prepay question
Billing can be pay-as-you-go or prepaid for a set number of periods. Prepaid annual plans are usually sold as a retention tactic. They are better understood as financing. The customer is lending you a year of working capital at the cost of a discount, and you are agreeing to deliver twelve months of service you have not yet performed. Deferred revenue is a liability wearing a revenue costume. It looks like a strong bank balance and it is partly an obligation.
Push annual prepay when your delivery cost is front-loaded and your retention is already proven. Avoid it when you are still learning what the offer should be, because refunds on a year of unearned service will hurt more than the monthly churn you were trying to avoid. And note that long billing cycles are no longer the low-maintenance option they once were: card network and jurisdictional rules now attach reminder obligations to them, which is the next point.
Compliance is no longer a legal afterthought. It is part of the model
This is where almost every ranking article on this topic is out of date, and the gap matters because it invalidates a retention tactic that older playbooks openly recommended.
In October 2024 the FTC announced a final “click-to-cancel” rule covering recurring subscriptions and negative option programs, with most provisions set to take effect 180 days after publication in the Federal Register. The rule prohibits failing to clearly disclose material terms, requires express informed consent before charging, and requires a simple cancellation mechanism that immediately stops further charges. Rules like this can be challenged and amended, so confirm the current status with counsel before you rely on any specific deadline. The direction of travel, though, is not in doubt, because the card networks arrived at the same place independently.
Visa requires merchants running free trials or introductory promotions to notify the cardholder at least seven days before the promotional period expires, and to provide a simple cancellation mechanism. Mastercard requires subscription terms to be displayed clearly and prominently at the point where payment credentials are captured, and for billing frequencies of six months or longer, an electronic reminder at least seven days before the next billing date. Billing platforms have already absorbed this into their products. Paddle changed its checkout in 2025 so that customers in California must tick a box agreeing to subsequent recurring charges, and its renewal notification system now varies by customer location, sending reminders seven or thirty days ahead depending on jurisdiction.
The design implication is blunt. Cancellation friction is finished as a retention strategy. Anything you were planning to retain through a hidden cancel button, a phone-only cancellation line, or a renewal the customer forgets about is revenue you should assume you will lose.
There is an upside to accepting this early. Treat the cancel flow as a product surface rather than a leak. Offer a pause, a skip, a downgrade, a smaller plan. A customer who pauses for two months is worth far more than one who cancels because pausing was not an option. And a model that survives a one-click cancel button is a model you can actually trust your forecast to.
A large share of your churn will not be a decision
Before you build win-back campaigns, check whether the customers you lost meant to leave. ProfitWell by Paddle reports that between 20% and 40% of users churn because they forgot to update card details, or because a payment failed and was never retried.
That is a plumbing problem, not a product problem, and the fixes are unglamorous: automatic card updating, intelligent retry logic on failed charges, dunning emails that reach a real inbox. Stripe states that its card account updater automatically refreshed more than two million expired or replaced cards for Postmates, which it translates to $60 million in revenue, and that its retry system recovered more than 200,000 failed payments worth $3 million. Those are vendor-reported figures from a company selling the tooling, so treat the magnitudes as indicative rather than typical. The underlying point stands on its own logic: a card expires every few years for every customer you have, and if nothing in your system handles that event, your churn number is partly fiction.
This matters more now because acquisition has got harder. Recurly’s 2025 report found subscriber acquisition rates dropped to 2.8% in 2024, down from 4.1% in 2021, and that 58% of consumers subscribed to between one and three services in the past year. People still buy subscriptions. They just hold a small number of them at once, which means you are competing for a slot rather than for attention. Keeping a paying customer whose card simply expired is the cheapest growth available, and it compounds the same way trust lowers your acquisition costs over time.
Put it in the plan properly
The SBA’s guidance on lean business plans asks you to define your revenue streams explicitly, and lists membership fees as one of its own examples. It also says businesses with multiple revenue streams should list them all. That second instruction is worth dwelling on, because it quietly contradicts the idea that you should go all-in on subscriptions. A mixed model is not a failure of nerve. Project work plus retainers, one-off sales plus replenishment plans, is often the structurally sounder business.
On projections, the SBA recommends a five-year financial outlook with the first year broken down quarterly or even monthly. For a recurring model, insist on monthly. Quarterly projections hide the specific thing that kills subscription businesses in year one, which is the gap between when you spend to acquire a customer and when that customer’s payments have covered the cost. Model it as a monthly cash line, not an annual average.
Three numbers I would not run this model without:
- Months to recover acquisition cost. If a customer takes eleven months to pay back what you spent to win them and your median customer stays nine months, growth makes you poorer. Faster growth makes you poorer faster.
- Cash collected versus revenue earned. With prepayments these diverge, and the bank balance will flatter you. This is the recurring-revenue version of the broader difference between revenue growth and profitable growth.
- Concentration. What percentage of recurring revenue sits with your largest three accounts? Above roughly a third, you do not have predictable revenue. You have three relationships and a spreadsheet.
The transition is a cash flow problem, not a marketing problem
Established businesses moving from one-off sales to recurring revenue hit the same trough. A $4,000 project becomes a $500 monthly plan, which is better after ten months and worse for the first eight. If you convert too much of the book at once, you can starve a profitable business into a liquidity crisis while every chart points upward.
A few things make the crossing safer. Keep selling one-off work during the transition, and fund the recurring build from it rather than replacing it. Start with the part of your delivery that already repeats naturally, usually the aftercare, the monitoring, the maintenance, the part customers keep calling about anyway. That is often also the most profitable part of your business, which makes it the right thing to protect with a contract. Offer the plan to existing customers first, since they already know what you deliver and need no persuading about whether the value is real.
And price the plan against the annual value of the relationship rather than against the monthly cost of delivery. A plan priced only to cover this month’s work removes your ability to absorb the heavy months, and heavy months always come. The same offer discipline applies here as anywhere else, which is why it is worth revisiting what actually makes an offer convert before you rebuild your pricing page around it.
The businesses that get this right are rarely the ones with the cleverest pricing. They are the ones whose customers would have come back anyway, and who simply made returning the default. Recurring revenue is best understood as a way of formalising a relationship that already exists. Where no such relationship exists, no billing cycle will create one.
