July 8, 2026
The subscription dilemma: when recurring revenue starts eating the product

Every product person eventually hears that recurring revenue is the goal worth chasing. Investors reward it, boards forecast against it, and a subscription turns a one-time buyer into an annuity. The trouble starts with that same math: the moment your revenue depends on people not leaving, your incentives stop aligning with theirs.
I work in payments, so I see this from the billing side. A subscription is a machine you bolt onto the product, and like any machine, it shapes the thing it's attached to.
How subscriptions degrade the product
The degradation is rarely a single decision. It's a slow drift, and it usually shows up as some combination of these:
Paywalling what used to be free. Features ship free to build habit, then get moved behind the plan once you're hooked. The product didn't get better — the wall moved.
Artificial scarcity. Limits that exist for no technical reason: five exports a month, three projects, ten prompts, engineered purely to make you feel the ceiling and reach for your card.
Nagware. Interstitials, badges, and 'you're missing out' banners that treat the free tier as a punishment to escape rather than a product to enjoy.
The hostage clause. 'Cancel and lose your history, your data, your streak.' The product stops being something you use and becomes something that holds your stuff.
The cancellation maze. The purest tell. Signing up took one click; leaving takes a gauntlet of discounts, guilt, and fake errors.
None of these makes the product better. Every one of them makes the metrics better for a quarter or two. That's the dilemma: the same instrument that funds a great product tempts you, every single sprint, to make it a little worse.
Feel the anti-pattern yourself
Reading about a dark pattern is one thing. Here's the cancellation maze made playable. Try to leave the imaginary plan — no tricks, there really is an exit — and notice how quickly 'valued customer' turns into 'flight risk to be neutralized.'
✦ Try to cancel
The subscription you can (barely) leave
Go ahead — cancel the imaginary $12/month plan below. This is every hostile retention flow you've ever met, stitched into one.
It's funny until you remember it's real. And increasingly, it's illegal.
Regulators caught on. The US FTC's click-to-cancel rule and the EU and California auto-renewal laws now require that cancelling be at least as easy as signing up. If your retention strategy only works because leaving is hard, you don't have a retention strategy; you have a liability.
Subscriptions done right
None of this means subscriptions are evil. The good ones share a simple trait: the value is genuinely continuous, so charging continuously is honest. A few patterns worth copying:
Charge for ongoing value, not for a hostage. Netflix and Spotify add new content every week; the subscription buys a stream of value that keeps flowing. You're paying for this month, not for access to last month's stuff.
Keep making the product better. Figma and Notion earn the renewal by shipping: the thing you're subscribed to in a year is meaningfully better than the one you signed up for. The wall doesn't move inward; the value moves outward.
Make leaving painless, and win on merit. The confident products (1Password, Basecamp) let you cancel in two clicks and export everything. It signals they expect the product, not the exit maze, to keep you.
Grandfather your early believers. Honoring old pricing for the people who bought in early buys more loyalty than any '50% off, please stay' popup ever will.
The billing surface counts too. Dunning, how you handle a failed payment, can be a quiet, respectful 'your card expired, here's a link,' or it can be a locked account and a guilt email. Same event, opposite message about who you think the customer is.
The test I'd apply
Before adding a subscription, or a limit, or a retention flow, I'd ask one question: would this survive a world where cancelling is one click and free tiers are generous? If the honest answer is 'no, people would leave,' the problem isn't the cancel button; it's that you're relying on friction instead of value.
Build the version that would keep the customer even when the door is wide open. That's the only subscription worth having and, conveniently, the only one that's still legal to run.
Get new posts by email
Occasional writing on payments, fintech, and building products. No spam, unsubscribe anytime.