Somewhere a green cartoon owl is nervously checking the news.

On November 11, 2026, the European Commission proposed the Digital Fairness Act. It's a draft regulation aimed at the tricks apps use to keep you scrolling, paying and forgetting to cancel. It isn't law yet: the European Parliament and the Council still have to agree on the text. But it tells you clearly where product design in Europe is heading.

I've been building web products for more than 13 years, mostly in PHP and Symfony, and I've launched startups that failed and a few that worked. So I read this as the person who would have to build the changes, not as a lawyer.

What the draft actually says

The Commission's own summary promises an end to "free trials turning into subscription traps", subscriptions that are "easier to cancel", a ban on fake "almost sold out" claims, and "the option for us to turn off addictive design features".

The more specific parts come from PPC Land's walk-through of the draft text, so treat the article numbers as reported:

  • Article 17, addictive design. If your product encourages compulsive use, users must be able to switch off autoplay, notifications unrelated to what they're doing, and "rewards for engagement at regular times". That last one is a polite legal way of saying streaks.
  • Article 20, free trials. Before a trial ends, you must send a reminder with the end date, the price and how to cancel. The paid contract continues only if the person expressly agrees.
  • Article 23, the cancel button. If people can sign up online, they must be able to leave online, through a button labelled "terminate the contract here", a confirmation step, and an acknowledgement within a week.
  • Minors. Virtual currencies are reportedly disabled by default for kids, and loot boxes are off the table for them.
  • Fines. Up to 6% of worldwide annual turnover.
  • Timing. If adopted, it applies 12 months after it enters into force.

Note what's missing: a ban. Autoplay and streaks stay. People just get a switch.

Why I mostly like it

A confession: in every product team, the cancel flow is the ticket nobody fights over. It doesn't bring revenue, so it slides to "next sprint" forever. I've shipped worse things than a hidden cancel link, and I'm not proud of any of them.

Here's the plain-language test I use now. If your revenue chart drops the moment cancelling becomes easy, you didn't have customers. You had people who forgot about you. That's the gym-membership model, and it's a bad foundation for a startup. Churn you can see is churn you can fix.

The trial reminder is the same story. Think of a hotel minibar that texts you before charging for the peanuts. Annoying for the hotel, and very fair to you.

Where I'm skeptical

Two honest worries.

First, compliance costs aren't equal. A big platform has a legal team and a design system. A three-person startup has one founder reading regulation PDFs at 2am instead of talking to customers. Finance Watch, quoted by PPC Land, also argues that the opt-out model leaves the burden on consumers, so even consumer groups don't think this is perfect.

Second, some people really do like their streaks. A streak got plenty of people through a language course they'd otherwise have quit by week two. Making it optional seems right. Treating every engagement feature as evil would be wrong.

PPC Land also flags a loophole: the free-trial definition covers trials with no payment, so a €1 "trial" might slip through. Expect some clever pricing pages.

What I'd check in my own product now

You don't need to wait for the final text. These are cheap and good for trust anyway:

  1. Can someone cancel in the same place, with the same effort, as they signed up? If cancelling needs an email to support, fix that first.
  2. Do you remind people before a trial converts? One email three days before, with the date and price. It costs almost nothing.
  3. Do your notifications earn their place? List every push and email. If one exists only to pull people back, give it an off switch.
  4. Any fake urgency? Countdown timers that reset and "only 2 left" labels that are never true should go.

If you sell to people in the EU, a 12-month runway sounds long. In my experience it isn't, especially for billing code nobody wants to touch.

The question

I think an off switch, not a ban, is the fair middle. Users get control, and products that are actually good keep their users without the tricks.

So, would you switch your streaks off, or keep them on? Why?