Easy to Leave

1. Make cancellation truly self-serve. Customers should be able to cancel without phone calls, retention specialists, or department transfers — one click, one confirmation, done. Anything else is a deliberate choice by the company to make leaving painful.
2. Let people take their data with them in a human-usable form. Basecamp exports data as a browsable, mini-website-style HTML archive rather than a raw CSV dump, so customers can actually use what they leave with.
3. Offer a 'pause' as a humane alternative to cancellation. 37signals lets customers pay a heavily reduced fee (around five dollars a month) for up to a year to keep all their data, people, and projects warm — useful for customers who love the product but temporarily have no work.
4. Don't trade your reputation for an A/B-tested uplift. Discount-to-stay offers, last-minute coupons, and forced phone calls might squeeze out a small percentage gain, but they ferment lasting anger and kill the referrals every business depends on.
5. Watch the leaky bucket. Jeff Bezos's early question to 37signals was whether customers were quietly disappearing. Pouring more marketing in doesn't matter if you're losing them out the bottom because the experience makes them unwilling to recommend you.
6. Price fairly and publicly — skip the coupons. 37signals avoids coupon codes and secret discounts because they make paying customers feel like suckers. A single fair price published openly is more trustworthy than a price you have to negotiate or scrounge the internet to get.
7. Eliminate haggling wherever you can. Tesla's no-negotiation, app-based car buying and 37signals' published, level-based salaries both remove the 'am I being taken advantage of?' question — and the relief that creates builds deep loyalty.
8. Offer fixed pricing alongside per-seat pricing. Basecamp has long offered a fixed-price plan in addition to per-user pricing, giving customers a predictable, no-haggle option that's rare in SaaS.
9. Don't trial-trap with credit cards up front. 37signals stopped requiring credit cards for trials roughly 15-20 years ago. Free signup with no card builds trust; surprise-billing customers who forget to cancel does the opposite.

The Philosophy: Easy On, Easy Off

Jason frames the whole discussion as a golden-rule problem. He doesn't like being stuck in software, contracts, or subscriptions, so he refuses to put his own customers in that position. Outside of situations with genuinely large capital expenditures, there is no technical reason cancellation should be hard — closing an account is, as David puts it, just a flag in a database. When a company makes leaving difficult, that is a deliberate choice, and it's not one 37signals is willing to make.

The bar 37signals holds itself to: a customer should be able to cancel without talking to anyone, without being routed to a retention specialist, without being 'talked off a ledge.' Click cancel, maybe confirm because it's a meaningful move, and you're done. That has been their approach across every variation of their products.

Getting Your Data Out

Cancellation is only half the story; the other half is what happens to your data. Jason notes this is genuinely hard in proprietary systems. 37signals has experimented with several approaches over the years and currently settled on a nicely formatted HTML export for Basecamp — essentially a mini-website you can browse — because it prioritizes humans being able to read and use the data, rather than dumping a giant CSV file that's hard to look at or import elsewhere.

Jason concedes it may not be the optimal format, but it reflects the principle: portability should feel respectful, not punitive. Some data types are naturally more portable than others — email and contacts move more easily than projects in a proprietary tool — but the goal is consistent: make the exit useful.

Pause as a Humane Alternative

Through their optional cancellation survey, 37signals kept hearing the same thing: 'I love Basecamp, but we don't have any projects right now.' Cancelling in that case is punitive — customers lose access to their projects and people, and even with an export, they can't really pick up where they left off when work returns.

Their solution was a pause option: a significantly reduced fee (Jason recalls around $5/month) for up to roughly a year. The account stays intact — data, people, projects, all warm and ready. When the customer comes back, nothing is lost. It's a way of recognizing that 'I'm leaving' and 'I don't need this right now' are very different signals, and treating them differently.

Why Retention Games Are a Bad Trade

David ties the dark-pattern epidemic to A/B testing culture. Someone is tasked with squeezing the last couple of percent out of a cancellation flow, and in isolation a discount offer or extra friction looks like free money. What's missing from that calculation is that people talk, recommend, and — when wronged — carry rage for years or decades. Every business is a referral business, so cementing animosity to hold a customer for one more month is almost always a bad trade.

The Adobe Cautionary Tale

David singles out Adobe as the worst offender he's seen. When customers try to cancel, they're not met with a discount offer but with what he calls a ransom: a bill for the remaining balance of an annual plan that was billed monthly — in his example, $172 owed at the moment of cancellation. He calls it 'the most perverted payment scheme' he's ever encountered and notes that the hatred this one program generates is unprecedented. Yet Adobe apparently keeps it, presumably because someone has decided the rage of 50,000 customers is worth a ~1.74% non-cancellation uplift.

The Financial Times Story

David subscribed online to the Financial Times from the US, then later realized he wasn't reading it and tried to cancel — only to discover he had to call a UK phone number within UK business hours. He remembers the incident seven years later and the lasting verdict it produced: he will never be a customer again. The asymmetry — happy to take a foreign customer online, but demanding they cancel as if they live next door — is what made the experience feel demeaning.

Discount-to-Stay Offers

Both Jason and David are skeptical even of the relatively benign 'half off if you stay' offers, which 37signals itself has used in a previous version of Basecamp. The problem, as Kimberly raises and Jason confirms, is that it tells the customer they've been overpaying — and produces a reluctant customer who didn't really want you anyway. Jason argues these tricks generate 'reluctant customers' rather than fans, and the math that says 15% will forget to cancel after the offer expires is, in his view, just a game he doesn't want to play.

The Leaky Bucket

David recounts that before Jeff Bezos invested in 37signals, the one thing he wanted to know was whether the business had a leaky bucket. Pouring marketing dollars in is meaningless if customers evaporate out the bottom because the product or experience leaves a bad taste. The way to keep that bucket tight is to leave a good impression on everyone — customers staying and customers going — so they'll recommend you regardless. Every business, they emphasize, is a referral business.

The Car-Buying Parallel: Tesla vs. Everyone Else

Jason is in the middle of trying to buy a Kia and the experience has become a vivid case study. He and David use it to extend the 'easy to leave' principle to 'easy to enter' — the buying experience itself can poison the relationship before it even starts.

What Tesla Got Right

Buying a Tesla, Jason recalls, is almost entirely app-based. You pick from roughly four models and a handful of simple choices — wheel size, interior color, exterior color, range. Packages don't cancel each other out the way they do at traditional dealers. The paperwork is digital, you sign with a button, they tell you when it's ready, and pickup involves walking up to a car with your name on it, pairing your phone, and driving away. There is no dealership experience at all. Once you've seen it work this way, Jason says, going back to the old model is painful.

The Kia Dance

Contrast that with Jason's current Kia experience: factory-installed paint and interior protection charges he didn't want and that supposedly can't be removed, an unitemized $500 'extra fees' line that turned out to be taxes and registration (so why bundle it?), multiple back-and-forths with a manager, the inevitable finance-guy step, and a looming three-hour Saturday at the dealership full of small-talk theater ('great color, great choice, my wife has one'). He's so worn down by the process that he's reluctant to actually complete the purchase he's already agreed on a price for. He suspects the implicit logic is sunk-cost psychology — the more invested you are, the less likely you walk.

Saturn and 'Kia Easy'

Jason brings up Saturn as a partial precedent: a no-haggle US brand whose prices were fixed, but where the rest of the dealership ritual — salesperson, paperwork, finance — remained intact. The price became un-negotiable but the bad parts didn't go away. He's now noticing similar 'easy buy' buttons on Kia's site, which appear to mean you skip negotiation but still go through the standard dealership funnel. He suspects these are aimed at people who simply don't want to haggle, but it's unclear whether they're paying more for the privilege — and either way, the worst parts of the experience remain.

Why Tesla Loyalty Is Different

David's observation: people who buy Teslas often convert into 'I will never buy another brand' customers. He doesn't hear that from GM owners. Some of that loyalty, he argues, comes from the relief of not being asked to be a tough negotiator — of not feeling like a sucker if you pay sticker. The historic dealer model in the US, partly preserved by lobbying that initially kept Tesla out of some states, treats car buying like a Turkish bazaar where the listed price is fiction and you need a magic 'real' number from the internet to avoid being taken. Removing that entire game is itself a feature.

Pricing, Coupons, and Trials at 37signals

37signals applies the same no-games philosophy to its own pricing. They don't run coupon codes (with a handful of historical exceptions) because coupons make the people paying full price feel like suckers and undermine the sense that the price is just the price. The published nonprofit discount and free Basecamp for schools are exceptions, but those are transparent, not secret backdoors. Jason concedes coupons probably work in some narrow sense, but asks 'work in what way?' — they may also be quietly eroding trust.

On trials, Kimberly notes how common — and how irritating — it is to be required to enter a credit card up front and then have to remember to cancel. Jason acknowledges 37signals did this in the early days, partly because of bank or processor requirements and partly because building a system to bill after the fact (with grace periods, payment links, etc.) was simply more complex. They abandoned that model 15-20 years ago. Today you sign up free, no card required, and only enter payment information if you decide to keep using the product.

Finally, Basecamp has long offered a fixed-price plan in addition to per-seat pricing — a rarity in SaaS, where per-seat (and increasingly consumption-based) pricing dominate. The fixed price is itself a kind of no-haggle promise: do the math, pick whichever option suits you, and don't worry about creeping costs. Jason believes 37signals should always offer something like that.

Applying No-Haggle Internally: Salaries

David extends the principle inside the company. Roughly 15 years ago, 37signals got rid of salary negotiation entirely. Everyone at the same level earns the same amount, and the salary is published right in the job posting. If it's not to your liking, you can apply elsewhere — but the haggling theater is gone.

He recalls hating the old annual-review dance of the 2000s, where every year became a Turkish-bazaar exchange about competitive offers and raises. He didn't want that to be his job and didn't want to give it to anyone else. When they first removed it, he half-wondered if they were doing something illegal — but of course paying everyone the same for the same level of work is perfectly fine, and most employees consider it a net benefit. It's the closest internal analog to the Tesla model: same level, same price, no games.