Subscription vs one-time: you are choosing which work you want.

Churn work or launch work. That is the decision.

The comparison is almost always made on revenue shape — a sawtooth against a rising line — and that framing hides the thing that actually determines which one you can sustain.

Sell once and every month starts empty. The work is finding buyers: a launch, a campaign, a webinar, a partner send, whatever fills the top of the month. It is visible, bounded and repeatable, and when it goes badly you know within a week.

Sell a subscription and every month starts with last month’s customers still in it. The work is keeping them: delivering something they would miss, noticing the ones who have gone quiet, and shipping in the eleventh month with the same care as the first. It is invisible, unbounded, and when it goes badly you find out six weeks later, from a cancellation whose cause happened in month three.

Both are real jobs. Neither is passive income, and the choice between them is mostly a question about you: which of those two rhythms you can actually run for two years without resenting it. The rest of this guide is about making that choice honestly and then implementing it without damaging the people who pay you.

A payment plan is not recurring revenue. It is one price, collected slowly.

This distinction sounds pedantic until the first refund request arrives, at which point it turns out to have been the whole argument.

A payment plan splits a fixed price into instalments. The customer is buying the thing outright; the schedule is a convenience. Three payments of one-third, and at the end they own it — access does not depend on anything continuing, because nothing is continuing. Your obligation is delivery, once.

A subscription is rent. The customer pays for a period and receives that period. When payment stops, access stops, because the thing being paid for was never ownership. Your obligation renews with every charge.

Both models produce a card being charged monthly, which is why they get blurred in the copy, and buyers resolve ambiguity in the direction that favours them. Somebody who thought they were paying off a course and discovers they were renting one does not write to say they misread the page — they ask for their money back, and they are not entirely wrong to.

The related trap is the instalment plan that revokes access if a payment fails. That is a reasonable protection and it needs to be said out loud on the page, because it is the detail that makes a payment plan behave like a subscription at the worst possible moment.

Ongoing delivery justifies a subscription. Ongoing billing does not.

This is the whole test, and it is worth applying before any of the arithmetic, because a subscription the customer does not believe in churns regardless of what the model says it should do.

Three questions, answered honestly about the product you actually have rather than the one you intend to have by December.

1. 01Is there something each period a customer would notice missing? Not “could there be” — is there, this month, something that arrives, changes, or happens. If you have to think about it, the customer has already noticed.
2. 02Does the value decay if you stop working? A community without a host degrades in weeks. A recorded course does not degrade at all, which is a compliment to the course and an argument against renting it.
3. 03Would a customer who stops paying reasonably expect to lose access? This is the fairness question, and it is the one that predicts disputes. If losing access would feel like a punishment rather than a consequence, the subscription framing is wrong.

Three yes answers and a subscription describes the truth. Two, and you probably have a one-time product with something recurring beside it — which is the hybrid, and it is a good place to be. One or none, and charging monthly is a decision to bill for something that is not happening.

Why the wrong answer is expensive rather than merely wrong

A subscription without ongoing delivery does not fail immediately; it fails on a delay. The first two or three months look fine because people are still consuming what they bought. Then the consumption ends, the charge does not, and the cancellations arrive together — usually alongside a handful of refund requests for the months where nothing happened, and those requests are reasonable. You are then choosing between refunding revenue you have spent and defending a position you know is weak.

The differences that matter before you have customers.

Not a scorecard — there is no winner. A list of the axes on which the two models genuinely diverge, so that whichever you choose, you choose it knowing what comes with it.

Two rows deserve a second read. “Where the work is” is the one people underestimate before choosing, and “where it breaks” is the one they underestimate afterwards. A subscription that keeps billing after delivery has quietly stopped is not a bad quarter — it is the thing that ends the relationship with an entire customer base at once.

Most honest answers are a one-time core with something recurring beside it.

The two models are usually presented as a fork. In practice the products that survive tend to charge once for the finished thing and monthly for the part that genuinely continues — because that is what is actually true about them.

Three rules that keep a hybrid honest

1. 01The core has to be complete without the layer. Someone who buys the course and never joins the community should get everything the sales page promised. If they do not, the community is not a layer — it is a missing chapter with a subscription attached.
2. 02The layer has to be cancellable without breaking the core. Cancelling the recurring part must never remove access to the thing they bought outright. This sounds obvious and is the most common way hybrids go wrong, usually by accident in the access configuration.
3. 03Never move something from the core into the layer after the sale. Taking a feature people already bought and putting it behind a monthly fee is the same betrayal as retroactive subscriptions, in a smaller costume. New things can go in the layer; existing things cannot move there.

The hybrid also solves a sequencing problem. You can sell the one-time core first, watch what buyers ask for next, and build the recurring layer out of the requests — which means the subscription launches with a delivery obligation you already know you can meet. That is a considerably better position than committing to monthly delivery on a guess.

Two lines, and the one people forget to draw.

Everything below is arithmetic on invented assumptions — it is a worked example, not a benchmark and not a forecast. Replace every number with yours; the shape of the answer is the part worth keeping.

What the table is actually showing

• →The subscription is behind for a long time. On these assumptions it takes until somewhere in the third year for cumulative cash to catch up. That gap is the real cost of choosing recurring, and it is why subscriptions are usually a second product rather than a first one.
• →The base settles, and lower than people expect. Customers added each month divided by the share who leave each month gives the size it settles at: twenty divided by five per cent is four hundred. The subscription in this example is not going to two thousand customers — it is going to four hundred, slowly, and then stopping.
• →The one-time line never rises on its own. Six thousand in month one and six thousand in month twenty-four, because nothing accumulates. Everything that grows a one-time business — audience, proof, referrals, a second product — is work that sits outside this arithmetic.
• →Month twenty-five is the honest comparison. One model opens it at zero with twenty buyers to find. The other opens it with several hundred people who need a reason to stay. Which of those you would rather wake up to is a better decision input than either revenue line.

Rebuild this with your own price, your own monthly acquisition and — the hard one — an honest cancellation rate. If you have no cancellation history, run the arithmetic twice at two very different rates and notice how far apart the answers are. That distance is a fair measure of how much you currently know.

An easy exit builds the brand that makes the next sale.

Cancellation friction does not retain customers. It delays them by a few weeks and converts a neutral event into a story — and the story outlives the few weeks by years.

The reasoning is not only ethical. Somebody who cancels cleanly stays on your list, keeps reading, recommends you to someone whose situation fits better, and comes back when theirs changes. Somebody who had to fight to leave does none of those, and tells the story in the exact places your next buyers are reading. The saved fortnight of revenue is the most expensive money in the business.

1. 01Cancelling happens in the account, not in an inbox. If a customer has to email a person to stop paying you, the cancellation is not a cancellation — it is a negotiation they did not ask for. One clearly labelled control, reachable without a search.
2. 02Confirm it in writing, immediately. The confirmation states what was cancelled, the date the last charge already happened, the date access ends, and what happens to anything they created or downloaded. Silence after a cancellation is what produces disputes a fortnight later.
3. 03Let them keep what they have already paid for. A period paid for is a period delivered. Ending access the instant somebody cancels, mid-period, reads as punishment and is remembered as one — for a saving of a few weeks of access that cost you nothing.
4. 04One save offer, at most, and never as a condition. A single honest alternative — pause, downgrade, or a smaller plan — is a service. A sequence of offers standing between the customer and the cancel button is a maze, and everyone recognises a maze.
5. 05Ask one question, after the cancellation has completed. Not before, and not as a required field. The answers from people who have already left are the most honest feedback you will ever receive about the product, precisely because there is nothing left to be polite about.
6. 06Stop the campaigns that assume they are still a customer. Nothing undoes a graceful cancellation faster than next Tuesday’s member newsletter, or a sales sequence pitching the thing they just left. Cancelling has to change what they receive, not just what they are billed.

What to do with the people who leave

Move them to a list that treats them as former customers rather than lapsed leads. They know the product, they know your voice, and they are the warmest audience for the next thing you make — provided the last interaction was a graceful one. Re-engagement campaigns work on that group and fail completely on the group that had to email three times to stop being charged.

Recurring billing is an operations problem wearing a pricing costume.

The decision is usually made on the revenue model and paid for in the operations. These are the six recurring costs that do not appear in any comparison of the two lines.

1. 01Cards fail, and they fail silently. Expiry, replacement cards, banks declining a foreign charge. A failed payment is a customer you are still delivering to and no longer being paid by, and unless something chases it, it looks exactly like a cancellation you never noticed.
2. 02Access has to change on the right day. Every subscription has an end state, and somebody has to make sure the product agrees with the billing. Access that outlives the subscription is revenue quietly leaking; access that dies before the paid period ends is a support ticket with a grievance attached.
3. 03The month has to ship whether or not you feel like it. This is the cost nobody prices. A subscription is a standing deadline, and the eleventh month is the one that decides whether you have a business or an obligation you resent.
4. 04Upgrades, downgrades and pauses all have arithmetic. Proration, the timing of the next charge, what happens to a partly used period. Decide the answers once, write them down in plain language, and apply them the same way every time — inconsistency here is what turns a billing question into a trust question.
5. 05Refunds get more complicated, not less. A one-time refund is one decision. A subscription refund asks how many periods, and whether the customer used the thing during them. Write the answer into the policy before the first person asks.
6. 06Someone has to read the cancellation reasons. They arrive one at a time and mean nothing individually. Read fifty together and the pattern is usually a single fixable thing — an onboarding step, a missing feature, a promise the product does not keep in month three.

None of these are reasons not to run a subscription. They are the reasons to decide the answers before the first customer rather than during the first crisis — every one of them turns into a trust question if it is improvised, and trust questions are settled against you by default.

You can change the model. You cannot change it retroactively.

Switching is normal and usually a sign the business learned something. What damages people is not the change — it is discovering the change through a charge, or finding that something they owned has become something they rent.

1. 01One-time to subscription: grandfather permanently, and say so first. People who bought a thing outright bought it outright. Converting their purchase into a rental retroactively is the single fastest way to convert customers into people who warn other people. Announce the change before it happens, state in writing that existing buyers keep what they bought for as long as the product exists, and apply the new model only to new purchases.
2. 02Subscription to one-time: honour the period, then stop billing. Finish the periods people have already paid for, cancel the recurring charge yourself rather than making them do it, and tell them exactly what they now own. Offer the one-time price at a discount to anyone who has been paying — they funded the product that is now being sold as finished, and they will notice if that goes unacknowledged.
3. 03Raising the price: new customers first, existing ones with notice. The cleanest version is that existing subscribers keep their price indefinitely and the new price applies to new sign-ups. If that is not survivable, give at least one full billing cycle of written notice, state the new amount and the date, and make cancelling before it takes effect completely trivial. A price rise that arrives as a surprise charge is a chargeback with extra steps.
4. 04Retiring a subscription entirely: give the exit before you need it. If you are closing the thing, say when, stop taking new sign-ups immediately, keep delivering until the stated date, and give people whatever is portable — exports, downloads, recordings. A product that ends well is remembered better than one that limps.
5. 05Whatever the change, write it once and send it to everyone. The same message, to every affected customer, before they hear it anywhere else. Segmented explanations of a pricing change tend to become inconsistent explanations, and inconsistency is what people screenshot.

If fewer than eight of those ten are true, the subscription is not ready — and the honest interim move is usually the hybrid: sell the finished thing once, build the recurring layer out of what buyers ask for next, and launch it when you already know what you are promising to deliver every month.