A service counter seen from the customer's side, with a queue barrier and signage marking the way in and nothing marking the way out.

I once reviewed the journey a company had designed for customers who wanted to leave, and realised nobody in the room had counted what it would cost the person leaving.

You've decided to go. Before the service even starts, it wants a stack of things from you — and it wants all of them now.

The Insight: Somebody Chose Whose Time This Costs

The closing journey I reviewed at a healthcare company — the path customers take when they want to leave — had been designed. Somebody had sat down and drawn it. Before it began, it asked for about ten pieces of information. Some of it was genuinely new. The rest it already had: contact details, the account number, the claim number, bank details, a void cheque.

I asked why it had to come that way. The answer was that this is how we do things here. Nobody in that room had costed the customer's time, and regulation meant nobody had to. I raised it. It came down to unwillingness: changing it now would have meant changing too much. I lost.

I've written before about forms that demand what a service already knows. This is that at a different moment, and the moment makes it legible. At the exit, from someone with nothing left to gain, re-asking stops being friction. It becomes a reading of whose time the company treated as free. The charge isn't the re-asking. It's that all of it had to arrive before anything could happen.

Moynihan, Herd and Harvey call this administrative burden — and their sharpest claim is that how the effort gets divided between institution and person is a choice somebody makes. They were writing about citizens and states; carrying it into a private company is my extension, not theirs.

Real-World Lens: The One We Can All See

Amazon called the way out Iliad. Reporting at the time said that was the internal name for the path out of a Prime subscription — after the poem about a ten-year war — and the FTC repeated the name when it sued in 2023. Amazon settled in September 2025 for $2.5bn, and the order requires it to offer cancellation "using the same method that consumers used to sign up."

A receding row of near-identical screens, each showing another offer to stay.

Amazon's exit has a court order behind it. The healthcare company's has my notes.

Under the Hood: What Is the Friction For?

I find it useful to separate two reasons an exit gets hard. The real cut is not who gains — the organisation gains either way, once as revenue kept and once as work avoided. It's what the friction is for, and whether anyone ever looked at it from the leaver's side.

Obstruction

Captive compliance

The friction is for

changing your mind

the organisation's own record

The leaver's side was

designed against

never considered

What shifts it

a rule

usually someone with standing to count their time

Some exits do earn their friction. Cochrane's review found that planning each hospital discharge for older patients individually probably shortens the stay by about three-quarters of a day across eleven trials, and modestly cuts readmissions across seventeen. But notice whose job it is. The planning is the hospital's. At the healthcare company, the file was the customer's.

So What?

Whose time absorbs the work? In that closing journey, the customer's — including the parts the company could have filled in itself.

Germany's law is unusually prescriptive — since 2022 it has dictated the cancel button's wording. A year in, a consumer group's automated sweep of 2,946 sites found only 42% complied. It's worth seeing what it covers: ongoing paid contracts you could have signed up to on a website. Not financial services.

❝

“The rules reach the exits that look like their entrances.”

Brazil's 2022 decree goes further for the services it regulates: cancellation through every channel you could have joined by, and no details demanded before you reach a person. So a rule can reach part of captive compliance. Though that one covers only federally-regulated services, and no one has published an evaluation of whether exits there got easier.

No such rule reached that room. I named the cost anyway; it changed nothing. My reading is that I had no standing to make anyone carry it — and that's the honest version of the fix.

So next time you leave something, notice two things: how much of what they demand they already have, and how much they want before anything happens at all. One tells you whose time they think is free. The other tells you whose convenience set the sequence.

Reply and tell me: what's the last thing you tried to leave, and what did it ask you for that it already had?

Next Wednesday: Why won't they take my word for it? — the proof of address, the doctor's note, the receipt you didn't keep. What a service is really asking for when it asks you to prove something.

Forwarded this? The Listening Loop pulls apart one invisible piece of service design every Wednesday.

See you next week.

Go deeper: Mills and colleagues' Dark patterns and sludge audits (open access) — it proposes that a good service should be as easy to leave as it is to join, and gives you a method for auditing one.

Sources: