Three steps, two interstitials, and a button that isn't the one you want
Subscribing took one click. Cancelling is "Step 2 of 3", interrupted by a modal asking if you'd rather change plans — where the retention button is styled and "No thanks" is not. Adobe paid $150M over this exact flow.
Facts cited
| DOJ/FTC settlement with Adobe | $150 million | $75M civil penalty + $75M in free services |
|---|---|---|
| Complaint filed | June 2024 | DOJ on referral from the FTC |
| What was alleged | hidden early-termination fees + obstructed cancellation | on the "annual, paid monthly" plan specifically |
| Global life expectancy | 73 years | World Bank, 2023 |
Assumptions low → high
| What we can't measure | low | likely | high |
|---|---|---|---|
| People holding at least one paid subscriptionStreaming, software, news, gyms, storage. An extrapolation, not a measurement. | 800M | 1.5B | 2.5B |
| Cancellation attempts per person per year | 1 | 2 | 4 |
| Seconds lost versus a one-click cancelMulti-step wizards, retention interstitials, mandatory surveys, and re-reading which button actually cancels. | 60s | 180s | 420s |
Who can fix it
- Owner
- Adobe (worked example) — and effectively every subscription business
- Channel
- Consumer protection regulators · FTC/DOJ · national consumer authorities
Signing up took one click and a stored payment method. Cancelling looks like this:
- “Step 2 of 3” — a wizard, for the act of stopping
- A mandatory feedback survey before you’re allowed to proceed
- An interstitial modal — “Can we help you switch products?” — injected mid-flow
- Asymmetric buttons: “Change my plan” gets a border and visual weight; “No thanks” gets none
- “Keep your plan” parked in the corner throughout, in case you waver
- And at the bottom, “Continue” — where continuing means continuing to cancel, which is momentarily ambiguous by design
Every one of those is individually defensible. Together they’re a maze, and the maze is the point.
This one has a court record
Most cases on this site are arguments. This one has a settlement.
In June 2024 the DOJ, on referral from the FTC, sued Adobe over precisely this: obscured early-termination fees and a cancellation process requiring “multiple steps online or repeated interactions with customer service representatives.” Adobe agreed to a $150 million settlement — $75M civil penalty plus $75M in services. Adobe denied wrongdoing.
The complaint singled out the “annual plan, paid monthly” structure — which is exactly the plan in this screenshot, at £32.66/mo, with one day of trial left. Miss the window and you’re not on a monthly plan you can leave; you’re in an annual commitment with a fee to escape it.
The principle: symmetry of effort
Leaving should cost what joining cost. If subscribing is one click from a logged-in state, then cancelling is one click from a logged-in state. Any gap between those two numbers is deliberate — nobody accidentally builds a three-step wizard — and the size of the gap is a fair measure of how much a company is willing to inconvenience you to keep your money.
Everything in the flow above exploits the same three levers:
- Steps — each one is another chance to give up.
- Interruption — a retention offer placed where you expected a confirm button.
- Visual hierarchy — the button they want is styled; the button you want is plain text. Your eye finds the emphasised one, and emphasis is doing the work a lock isn’t allowed to do.
The math
1.5B subscribers × 2 cancellations/yr × 180s = ~300 lifetimes / year
And unlike most cases here, the time is the smaller harm. The real cost is the subscriptions people fail to cancel — the charge that lands because the maze outlasted their patience. That’s what $150M was about, and money isn’t a unit this site measures.
The first case where “difficulty: none” is a lie
A cancel button is trivial to build. Technically this is the easiest fix on the site.
But every other case here is an accident — nobody wanted Chrome’s blank username field or Finder’s 1970 timestamps; they’re just nobody’s job to fix. This one is designed. The friction is the product working as intended, which is why it took a federal lawsuit rather than a bug report. So “difficulty: none” is true of the code and false of the organisation, and that gap is the whole story of this case.
The Adobe specifics are court-documented; the global sizing is not. The settlement, the filing date, the allegations and the plan structure are all cited. The number of people holding subscriptions and how often they cancel are extrapolations — hence the band spanning a factor of forty. The confident claims are the pattern and the legal record, not the headline figure.
Adobe’s current flow may also differ from the one described in the complaint, since the settlement requires changes. This case documents the screenshot as observed.
Difficulty to fix: none
One click in. Three steps, two interstitials and a survey out. The asymmetry isn’t an oversight — it is the feature, and it cost $150 million to be told so.
Read this case's source — _cases/018-obstructed-cancellation.md