DMA: Google Says EU Search Quality Fell, and Forked

Google has told Reuters that complying with European law produced “the largest reduction in quality of service” in Google Search’s 29-year history. Ten days earlier it had announced that one of its spam penalties would stop applying to anyone searching from inside the European Economic Area, while continuing to apply everywhere else. Put those together and the same query, typed by two people in two places, now runs against two different products with two different enforcement regimes behind them.

Google spent eleven months giving an unfalsifiable answer about traffic. It produced a precise one within days of having a regulator to name.

What did Google actually say about European search quality?

Nick Fox, a Google senior vice president, told Reuters that Digital Markets Act compliance caused the largest reduction in quality of service in the 29 years Google Search has existed.¹ The Digital Markets Act is European Union law that restricts how the largest technology platforms may favor their own services.

The specific changes give price-comparison sites such as Booking.com and Expedia more prominent placement, and they strip real-time pricing out of Google’s own travel carousels. Google also says an earlier round of the same compliance work already cut free, direct booking traffic to European businesses by 30%.¹

All of that is confirmed, in the sense that Google said it on the record. Whether the quality genuinely fell is a separate question that nobody outside Google can currently answer.

Why is Google’s precision about European search quality worth noticing?

The same executive gave a very different kind of answer eleven months earlier, and the contrast is the story. I wrote in Google’s billions-of-clicks claim that Fox was one of three Google voices repeating a stable weekly-clicks figure across eleven months, none of them ever attaching a baseline, a denominator or a methodology. Asked whether AI features were costing publishers traffic, Google offered a number that could not be checked.

Now the same executive supplies a superlative and a percentage. The reduction is the largest in 29 years. Google puts the fall in direct booking traffic at 30%.¹ Neither figure comes with a methodology either, but both are shaped to be quoted, and both point at a regulator.

My read, and this is inference rather than anything Google has conceded: the willingness to quantify tracks who absorbs the blame. Vague when the answer would implicate Google’s own product decisions, precise when the answer implicates Brussels. That pattern is worth holding in mind through November, when the next tranche of compliance reporting is due, because more numbers are coming and they will arrive with the same asymmetry.

What changed in spam enforcement, and for whom?

Google stopped applying manual actions under its site reputation abuse policy to results shown to people in the European Economic Area, starting August 30, 2026, following discussions with the European Commission.² Site reputation abuse is the practice of publishing low-value third-party content on a strong domain to borrow its ranking strength, sometimes called parasite SEO. A manual action is a human-issued penalty at Google, not an algorithmic one.

The details of the carve-out matter more than the headline. The manual action still registers in Search Console. It still affects rankings for everyone outside the European Economic Area. Inside that area, Google says affected sections may instead be separated algorithmically, so they rank independently over time.²

So this is a carve-out, never an amnesty. One penalty now produces two outcomes depending on where the searcher is sitting, and Google reached that arrangement through conversations with a regulator instead of through a search-quality rationale.

Why fork spam enforcement instead of dropping the policy?

Keeping the policy while suspending one region’s penalties preserves something Google would lose by repealing it. The rule still stands, the manual action still gets issued, and the record of the violation still appears in the site owner’s own console. Only the consequence changes, and only in one place.

Read the design charitably or cynically, and both readings are mine rather than anything Google explained. A suspended consequence stays reversible where a withdrawn policy does not, so enforcement can resume without Google publishing a word if the regulatory conversation moves. Meanwhile the algorithmic separation Google describes for the European Economic Area does much of the same work quietly. A section that ranks independently has lost exactly the borrowed authority the policy existed to deny it.

Sites operating in both regions inherit the awkward part. You can be simultaneously penalized and not penalized, with one interface reporting the first state and nothing reporting the second.

What does a forked index mean for a site owner?

A forked index means a single property can carry two ranking realities at once, and standard reporting will average them into one misleading line. A publisher penalized for site reputation abuse now keeps the penalty in the rest of the world while European searchers see something Google is handling differently. Any aggregate visibility number spanning both regions is mixing two regimes.

Search Console will show a manual action whose effect is now partial, and nothing in that interface tells you which share of your impressions it still suppresses. European traffic can move for reasons with no equivalent anywhere else, so a global rollup reads as noise. And a recovery in one region tells you nothing at all about the other.

For travel and price-comparison in particular, the placement changes Fox described are a competitive reshuffle rather than a quality event. If your competitors are aggregators, they gained prominence in Europe by regulation, not by out-optimizing you.

What can a site owner check about a forked index?

A site owner can check very little of this directly, and I would rather say so plainly than dress this piece up as an audit. This is desk analysis. I have no first-hand measurement behind it, and the client visibility data that would test it is confidential and cannot be published in any form, including in aggregate. Treat nothing below as a finding. It is a list of what a site owner is positioned to look at.

Segment your Search Console performance data by country and hold European markets apart from everything else, then keep them apart from now on. That separation costs nothing today and it is the only way a change like this becomes visible later. If you carry a manual action, note the date and treat any partial recovery as a regional artifact until proven otherwise.

Watch November, when the next round of compliance data is due. Assume Google reports a decline, since it has already trailed one. The thing to check is whether a methodology arrives alongside the number this time, or whether the figure comes bare in the shape the company has used all year.