How to build a subscription website

We make Drry, which is one of the places you could build this, so we are not a neutral narrator. The mechanics below are the same wherever you build, and where we describe our own product we say so. Here is the argument: a subscription website is a different business from a shop, not a shop with a recurring charge bolted on, and most of the expensive mistakes come from treating it as the second thing.

A one-off sale ends at the checkout. That is the whole shape of it: you persuade somebody, they pay, the transaction closes, and your next job is to find another person. A subscription starts at the checkout. The person who paid you today will decide again in thirty days, and again thirty days after that, and nothing about the sale you already made is evidence about either decision. Everything below follows from that one difference.

What changes when the same person pays every month

In a shop, marketing is the engine. Conversion rate, traffic, the offer, the checkout: get those right and the business works, because each sale is complete in itself. On a subscription website the equivalent engine is retention, and it is not a marketing activity at all. It is whether the thing you promised keeps turning up.

The practical consequence is that your best work happens after the sale rather than before it. A launch that brings 100 members into something with nothing happening inside it does not produce 100 members, it produces a refund queue and a quiet feed by month three. The same launch into something with a weekly rhythm produces a base you can build on for years. Same traffic, same price, same page: a different business, decided entirely by what happens in the weeks after the card is charged.

It also changes what "done" means for the product. A course can be finished. A subscription cannot, because the month somebody pays for has to contain something. That is not an argument for making more content forever, and treating it as one is how people burn out in a year: a live call, an answered question, a members-only thread that is actually busy, and access to the archive are all perfectly good answers. The requirement is that something arrives, not that it is large.

The three numbers that decide whether it works

A subscription business is legible with three numbers and almost illegible without them: what you charge, how long somebody stays, and what it costs you to get them. The monthly revenue figure everybody quotes is not one of the three. It is an output.

Now the part that surprises people. There are two obvious ways to make a member worth more, and they are not remotely the same size.

That is not an argument against pricing properly, and underpricing does real damage of its own. It is an argument about where the next hour goes. An afternoon spent on the first-week experience, the welcome sequence, or the reason to open the site in week six is worth more than an afternoon spent on the pricing page, and the pricing page is the one that feels productive. If you do want the pricing decision in detail, we wrote it up separately in how much to charge for a membership.

The third number is what it costs to get a member, and it is the one that decides whether you can grow on purpose rather than by luck. Say you spend $600 on ads and 10 people join: that is $60 a member. At a lifetime value of $375 the trade is comfortably profitable, roughly six to one. But the $375 arrives over a year, and the $600 left your account this week, so the number to watch alongside it is payback: at $30 a month, $60 takes two months to come back. A business with good lifetime value and slow payback can be profitable on paper and out of cash in practice, which is the specific way subscription businesses die while looking healthy.

One honest caveat on all three. You do not know your own cancellation rate for the first several months, and no number from an article, including ours, is a substitute for yours: it varies enormously by audience, price and promise. Plan with a range, write down what you assumed, and replace it with your own figure as soon as you have two or three cohorts to read. Annual billing is the same kind of "it depends": it improves cash and apparent retention and it trades away the monthly signal that tells you something is wrong.

What a subscription website has to do that a shop does not

This is the build list, and it is why a shop plus a recurring charge is not the same product. Four jobs, none of which a one-off checkout has to think about at all.

• Recurring billing, and recovering a failed one. Cards expire, get reissued after fraud, and decline for reasons that have nothing to do with intent. Some meaningful share of your cancellations will not be decisions at all, and the platform has to retry the charge, tell the member to update the card, and keep their access open while that is happening. A subscription site with no recovery path loses members who wanted to stay, silently.
• Gating access, and removing it cleanly. Access has to open when the payment succeeds and close when the paid period ends, not instantly the moment somebody cancels, which cheats them out of time they paid for, and not never, which is the more common bug. Check how a platform handles the in-between states too: a cancelled member serving out their month, a failed renewal being retried, a refund. Those states are where amateur setups leak.
• A reason to come back this month. A shop can sit still between launches. A subscription cannot, because every renewal is a fresh judgement about whether the last thirty days were worth the money. Something has to be new, live, or unanswered: a call, a thread, a drop, an event on a calendar people can see coming. If your site cannot tell a member what is happening next, they will find out by cancelling.
• A way to leave without asking you. Make cancellation self-serve and obvious. Forcing an email produces chargebacks, angry reviews and support work, and it does not keep anybody: it delays them and costs you the goodwill that would have brought them back next season. Cancellation rules also vary by market and have been moving in recent years, so check what applies where you sell. Even where nothing compels you, this one is simply better business.

There is a fifth job that is easy to miss because it is invisible at launch: knowing who is about to leave. A shop measures orders. A subscription website needs to see who has not logged in for a month, which cohort is leaking, and what the people who left had in common. You do not need this in week one. You will want it badly by month six.

What it costs to run one

Platform pricing in this market comes in three shapes, and the pricing page often does not make clear which one you are being offered: a flat monthly fee, a percentage of everything your members pay you, or both. On top of any of them sits card processing, which goes to the payment processor and which you pay whatever you build on.

The percentage is the one that grows with you. A flat fee is a fixed cost: it is the same whether you earn nothing or a great deal, so it gets cheaper as a share of revenue every month that goes well. A percentage does the opposite. It rises every time you add a member, every time you raise your price, and every time a member stays another month, which are precisely the three things this whole guide is about doing. It is also the line people skip while choosing, because a percentage is a number you have to multiply before it means anything.

One dated example of the pure percentage shape, because a qualitative description of it gets waved away. Patreon: Free to start. The cost is the percentage, not a subscription. Platform fee (new creators): 10% of memberships and one-off purchases. Verified September 2026 against Patreon's creator fees overview. That is a real trade rather than a trick, and it is the right trade for somebody genuinely unsure whether anybody will pay: nothing leaves your pocket until money moves. It is simply a trade whose cost arrives in your best months. We work the crossover arithmetic, the revenue at which a percentage overtakes a flat plan, in platform fees are a tax on growth.

Then there are the costs that are not on the pricing page at all, and they are the ones that surprise people in month four:

• Emailing your own members. Frequently a separate product priced by list size, which means the cost of telling people you published something grows with the audience you built.
• Seats and storage. An extra admin, or video beyond a modest allowance, is often metered. Video is the usual culprit, because a subscription library only ever gets bigger.
• Whose payment account it is. Not a line item, but the most expensive one. If the subscriptions live with the platform rather than with you, moving later means asking every member to re-enter a card, and the realistic outcome of that is losing a slice of your paying base. That is the difference between a bill and a lock-in, and we made the fuller argument in own your audience, don't rent it.

Our own shape, stated so it can be discounted appropriately: Drry charges a flat monthly subscription and 0% of what your members pay you, on every plan. Members pay through your own Stripe account, which is also why the percentage can be zero: we never hold the money, so there is nothing to take a cut of. There is a free plan that holds up to 100 members with no card, then $29 and $99 a month.

When a subscription is the wrong model

Sometimes the answer is that you should not build a subscription website, and it is worth saying plainly because everybody selling software in this market has an incentive not to. A subscription is a promise that value keeps arriving. If your value arrives once and completely, a subscription is a worse product than the thing you already have.

• The value is delivered once. A course somebody finishes, a template pack, an ebook, a single transformation with a definite end. Sell it once, at a price that reflects the outcome, and keep the email list. A recurring charge bolted onto a finished thing churns as fast as people work through it.
• You cannot answer the month-four question. What does a member get in month four that they did not get in month one? If nothing comes to mind, the subscription will leak faster than you can refill it, and no pricing or platform decision fixes that.
• The work does not fit your life. A subscription is a standing commitment to show up, most obviously if live calls are the draw. If the months where you are travelling, ill, or simply out of ideas would leave the site empty, be honest about that before a hundred people are paying for it.

A cohort program with a start and an end, sold repeatedly, is often the better business for exactly the people who think they want a subscription: the same revenue, a clean finish line, and no obligation to be interesting in February.

What to look for when you choose where to build it

Most of this market can host content and take a payment. The differences that matter are in the four jobs above and in what leaving costs, so ask about those instead of comparing feature lists.

Here is our own answer set, disclosed as such. On Drry: subscriptions live in your Stripe account; a failed renewal is retried while the member keeps access and gets a branded note asking them to update the card, with a heads-up to you; access is gated on a live entitlement and ends when the paid period does; members cancel and resume themselves; the platform takes 0% of member payments; email to your members is included on every plan, free included, with a monthly send allowance per plan rather than a separate bill sized by your list; and export is self-serve on every plan. Ask the same seven of everybody else on your list, and prefer whoever answers them plainly rather than whoever answers them best. If you want the field laid out side by side, our membership platform roundup quotes the figures each platform publishes, with the date we last checked each one.