A self-preferencing probe that ends without a finding of guilt
On August 3, 2026, the Netherlands' Authority for Consumers and Markets (ACM) published a draft decision proposing to make binding a set of commitments from bol, the country's largest online marketplace. The case began with a straightforward concern: does bol's Buy Box — the default "add to cart" slot shown when multiple sellers list the same product — quietly favor bol's own retail arm and preferred partners over the third-party sellers who list on its platform? The ACM's investigation also examined whether bol used non-public seller data, gathered by virtue of running the marketplace, to give its own sales operation a competitive edge (ACM press release).
The public comment period on the draft decision runs through September 13, 2026 (ACM draft decision page).
The case for scrutiny, stated fairly
The concern here is not manufactured. A Buy Box is not just a UI convenience — on most marketplaces it determines which seller actually gets the sale in the overwhelming majority of transactions, since few shoppers scroll past the default listing. When the platform operator is also a competing seller, as bol is through its own retail inventory, the ranking algorithm becomes a lever it can pull against the very merchants who depend on the platform for reach. Layer on data access — a marketplace operator sees every seller's pricing, sell-through, and demand signals — and the operator can, in principle, out-compete the businesses it hosts using information those businesses cannot see about each other. This is precisely the theory that underpinned the European Commission's 2017 Google Shopping decision, which fined Google more than €2 billion for favoring its own comparison-shopping results, and it is the animating logic behind the EU's Digital Markets Act gatekeeper self-preferencing ban. Given that bol holds, by a wide margin, the largest position among online marketplaces in the Netherlands, sellers with few realistic alternatives have limited leverage to negotiate around unfavorable ranking or data practices on their own. A regulator declining to look at this would be neglecting its job.
What bol actually agreed to
What's notable is how the case resolved. Bol committed to base Buy Box placement on performance metrics only, ending any structural edge for its own listings or favored partners, and it has already equalized eligibility for its promotional "Select Deals" program between itself and third-party sellers (eCommerce News NL). It pledged not to use non-public seller performance data to benefit its own sales activity. And it agreed to open up search analytics to sellers — search terms, click data, click-through rates, product views, and average click position — information sellers previously could not see about how customers actually find their listings. Notably, bol had already replaced its ranking's "performance score" with a broader "quality score" in June 2025, before the ACM's probe concluded, suggesting the platform was adjusting ahead of, not purely because of, formal pressure.
Crucially, the ACM did not levy a fine, and its own account of the case states plainly that the investigation "did not establish any violation of laws or regulations" — the commitments are voluntary undertakings the authority now proposes to make legally binding, closing the file without a formal infringement finding (ACM press release). Bol's chief legal officer, Jorn Palm, framed it as a transparency fix rather than a concession to wrongdoing: "Millions of customers and thousands of selling partners choose bol every day. That means it must be clear to everyone how our platform works and which rules apply" (bol's statement).
Why this is the right model, not a loophole
Critics of commitments procedures argue they let dominant firms buy their way out of accountability — no fine, no admission, no deterrent effect for the next platform tempted to self-preference. That's a fair worry in the abstract, and it is exactly why the Digital Markets Act backstops commitments with hard prohibitions and steep penalties for gatekeepers that don't comply.
But for a national regulator working outside DMA gatekeeper status, the ACM's approach here gets the sequencing right. Litigation over algorithmic ranking is slow, expensive, and often resolves years after the competitive harm — if any existed — has already reshaped the market. A negotiated, published, publicly consultable commitments decision that delivers concrete behavioral change (verifiable ranking criteria, real data-use limits, and new seller-facing analytics) inside months, with the threat of binding enforcement if bol reneges, is a faster and more proportionate remedy than a multi-year abuse-of-dominance case with an uncertain outcome. It also preserves the presumption that platforms haven't broken the law until a regulator actually proves it — a distinction commitments procedures, unlike settlements that bundle in fines, keep intact.
The test now is enforcement, not the deal itself. If ACM's binding order lacks teeth — vague performance-metric definitions, no independent audit of Buy Box logic, no penalty regime for backsliding — sellers will be back where they started. The September 13 consultation window is where that gets tested before the decision is final.