The subscription trap is real, and it's not just the customers caught in it I read the stories this week about people paying £500 before they realised they couldn't cancel. That's not a customer service failure. That's a business model that's been optimised so hard in one direction that it snapped. The subscription model itself isn't the problem. The execution is. I've sat across from people running subscription businesses who genuinely believe their churn numbers are healthy. Then I ask them how many cancellation requests get stuck in support queues, how many people email three times before giving up, or what percentage of their refund requests are legitimate versus people just trying to get their money back after they've already got the value. The answers are usually not great. Here's what I think: if you're running a subscription business and you're not immediately uncomfortable with how easy you've made cancellation, you're probably too aggressive. Not morally. Commercially. Because the government is about to make it a legal requirement anyway. The legal ground is shifting under your feet New laws are coming that let people cancel subscriptions with one click and get refunds without friction. This isn't a suggestion. This is happening. When Ofcom or whoever enforces this starts investigating, the businesses that get caught out won't be the ones that made cancellation easy. They'll be the ones that made it hard. But that's not actually why you should care. The real question: is your subscription model solving a real problem, or just extracting money? I need you to be honest here. Most subscription models fail not because the cancellation process is too easy. They fail because the product doesn't actually improve enough month on month to justify staying. You know what I see a lot? People who built something once, packaged it as a monthly subscription, and then just... watched it. They release updates every few months. They don't talk to customers who cancelled. They don't know why people leave. They just assume it's price or competition, when actually it's that the product hasn't changed since they signed up. I worked with a software business in London last year that had a 45% annual churn rate. They thought their problem was customer acquisition cost. It wasn't. It was that they'd built a tool that solved a specific problem in month one, and then nothing happened. No new features. No improvements. Just the same tool, same price, every month. We ran through their cancellation feedback. 60% of people said something like "it doesn't do enough anymore" or "we found something better". The other 40% didn't even bother giving feedback. They just left. They fixed the product roadmap. Churn dropped to 28% within six months. Not because they made cancellation harder. Because they gave people a reason to stay. Three questions to ask yourself right now First: would people keep paying for this if they had to actively re-decide every month? Not "would it be easier if they forgot to cancel". Would they actually choose to pay? Second: how much of your revenue comes from people who want to leave but can't work out how? I don't have a number for you because I don't have access to your support tickets. But if you're not tracking this, you're flying blind. Start tracking it tomorrow. Third: what's actually changing between month one and month twelve? If it's nothing, you don't have a subscription business. You have a payment collection scheme. Those don't last. The uncomfortable truth Some subscription models are fundamentally broken and should be killed. Not because the customer acquisition was wrong or the pricing was off. Because the product doesn't actually get better, and you can't force someone to pay for stagnation forever. I've recommended to clients that they stop running subscriptions and move to perpetual licences or one-time payments. Sometimes that's the right call. It's rare, but it happens. Usually it's because they've built something genuinely useful that solves a problem once, and then it's done. You don't need a gym membership for a dumbbell you own. But I've also seen people bail on subscription models too early because they didn't want to do the work of actually building something people want to keep paying for. That's the opposite mistake. What you should actually do If you're running a subscription business, don't panic about the new laws. They're good. They force you to be honest. Instead, look at your churn data. Really look at it. Not the headline number. The reasons. Why do people actually leave? If you don't know, start asking. Add one question to your cancellation flow: "What's the main reason you're leaving?" You'll get better feedback in two weeks than you've had in a year. Then ask yourself whether the product is actually getting better. Not incrementally. Genuinely better. If your roadmap for the next six months is mostly bug fixes and small optimisations, your subscription model is probably struggling because it should be. The businesses I see succeed with subscriptions are the ones that treat the model as a commitment to constant improvement, not a license to be lazy. They talk to customers. They ship things. They measure whether people are actually using the new stuff. They iterate. That's harder than just charging people every month and hoping they forget to cancel. But it's also the only way to build something that lasts. What to do this week Pull your cancellation feedback from the last three months and categorise it by reason. If you don't have structured feedback, add that one question to your cancellation flow on Monday. You need real data before you can make any decision about this. Look at your product roadmap for the next six months. Count how many items are genuinely new features versus maintenance and fixes. If it's less than 60% new work, that's a warning sign. Document this, because you'll need it for the next step. Pick three customers who cancelled in the last month and email them asking why. Not a survey. An actual conversation. Offer them 30 minutes. You'll learn more from three real conversations than you will from any amount of data analysis. Do this before you make any big decisions about your model.