Why Google's Universal Cart is a slow extinction risk for Central European price-comparison sites — and the one move that could turn them into the gateway for agents

Every few years someone declares that price-comparison sites are dead — and every few years Árukereső, Compari, and Ceneje keep right on sending traffic. Fifty million visits a month across Central Europe, with gross profit near an all-time high. Heureka, the group behind them, overtook eMAG on market share in Romania while halving its debt. Anyone who bet on the death of price comparison has been losing that bet for twenty years.

So let's give the optimists their due first. These aren't small local sites waiting to be rescued — they're a PPF-backed conglomerate that has been suing Google for a decade, and whose strategy teams thought all of this through long ago. The argument that follows isn't about their death; it's about the one structural shift worth catching in time. Now let me flip it.

What does Universal Cart mean through the eyes of a price-comparison site?

On 19 May 2026, at Google I/O, Google unveiled Universal Cart: a smart cart that compares offers from multiple merchants, checks that products are compatible, and closes the purchase in a few taps. Read that again through a competitor's eyes. Comparing offers from multiple merchants is precisely the entire job of a price-comparison site. Google has now built that job into Search and the Gemini app, as a native feature.

The details that set the scale: it launches in the United States this summer, there's no European date yet, and the brand remains the merchant of record — so it doesn't bypass the store. The store's front door, however, is exactly what it intends to bypass, and the price-comparison site is that front door. That's what makes the question structural rather than seasonal. This isn't the next core update, and it isn't next week's event either — it's a layer expanding over several quarters that slowly rewrites where buyer intent lands.

Set Universal Cart alongside AP2, the open Agent Payments Protocol, which Google announced in September 2025 and donated to the FIDO Alliance in April 2026. AP2 is a standalone open protocol, separate from Universal Cart — not part of the same product. It lets an AI agent make purchases on the buyer's behalf, with a single authorization, across multiple merchants at once — and in its newer form, even when the buyer isn't present at the moment of purchase. When search, comparison, and payment all happen inside a single agent, the landing page you would have had to open to compare prices stops being a destination. This isn't a drop in traffic — it's the disappearance of the very reason the site exists at all.

The lawsuit protects, but does it build? The DMA as a shield

In a Prague meeting room, the instinct will be to reach for the lawyers — and they'd be right. The DMA is a real shield, not a speech. Google is a designated gatekeeper. If, in Europe, it favors its own shopping surface over rival comparison sites, that's exactly the conduct that drew a €2.42 billion fine back in 2017, led to Heureka's own case at the Court of Justice of the European Union, and produced a record damages award for sister site Idealo at a Berlin court in November 2025. Forty-three price-comparison sites jointly petitioned the Commission under the DMA.

The lawsuit, though, only defends the old business. It doesn't build the new one. You can win damages for being cut out of the process while you are, in fact, being cut out of it. The mistake would be to treat the court as a strategy rather than as cover. The law wins the time; what gets built in that time is decided by a different question.

How does a price-comparison site become a supplier to agents?

The real move is to stop being a landing page and become a supplier. Agents need exactly what a price-comparison site already has, and what Google's Shopping Graph covers poorly: the complete, canonical offer graph of the long tail of Central European merchants, with real prices, real availability, real trust signals, in Hungarian, Romanian, and Slovenian. The agent picks the merchant whose data feed best answers the agent's question. So make that offer graph available as an API built for agents — an MCP endpoint a model can query in real time — and become the place ChatGPT, Gemini, Claude, and Perplexity turn to when they need to know what's actually available in this region, and what it costs right now.

Then build on AP2 — don't just sue the company that donated it. For twenty years, a price-comparison site has been the trust layer between buyers and thousands of small merchants. Channel that trust toward the agents: become the aggregator that spans the long tail of Central European stores with a single authorization and vouches for which merchant is genuine. Own the layer of completeness and trust where a U.S.-centric cart will be weak at the European launch. Trust isn't a side note: according to Bain & Company's November 2025 analysis, buyers trust a merchant's own agent roughly three times more than a third-party agent — and the same study puts total U.S. agentic commerce at $300–500 billion by 2030 (Bain & Company, 2025). And use DMA interoperability on the offensive — not just for damages over exclusion, but to demand mandatory, agent-side access, as the interoperable European alternative to the single gatekeeper's cart.

The data, the merchants, the trust, and the legal position are already in Heureka's hands. Nothing has to be invented. It just has to be redirected — from a person opening a browser, to an agent opening an API. This is the same logic as the difference between classic search and AI visibility: whoever can be cited stays visible, whoever can't, disappears. I lay out that difference point by point in SEO versus GEO.

And what about those without a graph? The uncomfortable lesson for the smaller players

Here comes the uncomfortable lesson for everyone smaller than Heureka — which means everyone reading this. If even a fifty-million-visit conglomerate has to transform the entire reason for its existence from a website into a data feed an agent can read, then the question isn't whether your store shows up on Google today. It's whether it exists at all in a form a buyer's agent can choose. Price-comparison sites at least own the graph; most companies have nothing an agent can read in four seconds — and no idea that this is now the only audience that decides.

This is the same move as when Google's AI Overviews reshuffled the top of the results page: whoever became a source stayed visible, whoever didn't was hidden behind the finished answer. I unpack the parallel — and what it means for a Hungarian company today — in the piece on Google's AI Overviews. Exactly what that kind of readability measures, from crawler access through structured data to off-site presence, I describe across seven dimensions, and you can follow the full scoring logic through on the methodology page.

The landing-page era is closing. The supplier era is opening. In Central Europe, price-comparison sites get to decide which one they live in — and on a smaller scale, so do you.

The winners of agentic commerce won't be the sites with the most visitors, but the ones an agent can read. If you're curious how your own company appears to such an agent today, I'll show you the starting picture with a free mini-check. Request a free mini-check.

Kindl Dávid · Published: June 13, 2026