Europe’s Digital Markets Act
The DMA has yet to change what customers choose
Brussels has shown it can fine a gatekeeper, and should now be held to whether anyone uses the rivals the law was written to help.
Booking.com took a clause out of its contracts in July 2024. The clause had stopped hotels from offering a lower price on their own websites. The contract was fixed, but the habit lingered: until September 2026, prices charged elsewhere were still weighed in deciding which hotels qualified for Booking Sponsored Benefit, a scheme in which Booking.com cuts the price shown to customers and covers the difference itself.5 The rule had been obeyed for two years. The market had to wait a little longer.
That is the Digital Markets Act (DMA) in miniature. Three years after the Commission named its first gatekeepers, there are seven of them, covering 23 services.1 The fines have arrived: €500m for Apple over anti-steering and €200m for Meta over the choice it gave users on personal data in April 2025,3 then €890m for Google on July 23rd 2026, split between favouring its own services in search (€460m) and obstructing steering on Google Play (€430m).2 That is €1.59bn in three decisions. We think Brussels was right to write the law and has now shown it will use it. It has not yet shown that anything changed for the customer, and that is the test it should be held to.
The Commission prefers to count compliance. Its first review of the law, due by May 3rd 2026, found the DMA “fit for purpose” and concluded that it was “too early for legislative changes”.6 The evidence offered is a list of things that now exist: choice screens, data portability tools, about 40 outside developers using gatekeepers’ data-transfer interfaces. Alternative browsers and search engines such as Ecosia, Qwant and DuckDuckGo have grown, particularly in France and Germany.6 Grown from what, and to what? A start-up that rises from half a per cent to three-quarters of one has grown. A compliance report can be a Potemkin village: a handsome frontage, and nobody asks who lives behind it.
Speed is the part Brussels controls, and the record is mixed. The Commission sent Alphabet preliminary findings on both search and Google Play on March 19th 2025.7 The fines came 16 months later. Meta, fined in April 2025, committed in December 2025 to offer Facebook and Instagram users in the EU a choice of less personalised ads from January 2026, which the Commission calls the first such choice on those networks.4 Even then, the Commission said it would then seek feedback and evidence on uptake. Uptake is exactly the word we would build the test around.
So we would have the Commission publish, now, the numbers by which it will be judged. For each gatekeeper service that is under the law: how many people use a rival, how many businesses sell through one, and what customers pay. The law requires a further review every three years, so the next falls due by May 2029.6 If baselines are published today, that review can be read against something other than the Commission’s own enthusiasm. If the figures have not moved, the DMA is an expensive way of producing paperwork. If they have, Brussels will have earned the credit it keeps claiming.
The strongest objection is that this is unfair. Customers are sticky, defaults are powerful, and a regulator cannot make people prefer a different phone or search engine. A law that opens a door has done its job whether or not anyone walks through. There is truth in that, and it argues for patience, not for silence. Brussels already takes credit when rivals grow; a regulator that claims the wins must accept the losses. Nor is it asking too much to say in advance what a reasonable result would look like. If the answer is “very little, for a decade”, voters and gatekeepers alike should be told so now. Some of the evidence is already uncomfortable: the Booking.com episode shows that a clause can disappear from a contract and survive in a ranking rule.
Hotels were freed to undercut Booking.com in July 2024. It would be nice to know how many have.