Trivago Q2 2026 Earnings: €168.4M Revenue, 21% Growth
Most CEOs are using AI to sack people politely. Trivago’s Johannes Thomas is trying to turn 600 staff into the output of 6,000—and that is a far harder job.
Most CEOs are using AI to sack people politely. Trivago’s Johannes Thomas is trying to turn 600 staff into the output of 6,000—and that is a far harder job.
That is the real leadership story in AI right now. Not which chatbot writes the prettiest meeting notes. Not who can announce the biggest AI budget with the straightest face. The question is whether a boss can redesign how work gets done before competitors do it for them.
The numbers say Trivago has earned the right to try
Trivago reported €168.4 million in second-quarter 2026 revenue, up 21% year-on-year. It was the company’s sixth consecutive quarter of double-digit growth. Management also raised its full-year adjusted EBITDA target to around €30 million and said it now expects mid-teens percentage revenue growth for 2026.
That matters because this is not a hot AI startup selling a dream to people with more optimism than financial discipline. Trivago is a travel-search business operating in a brutal neighbourhood. Google can answer travel queries. Booking Holdings has enormous scale. Expedia is its majority owner. Every travel platform is fighting for the same traveller, the same hotel inventory, and the same paid-search real estate.
In other words, this is the sort of company that should get squeezed into irrelevance.
Instead, Trivago generated €548.9 million in revenue in 2025, up 19% from €460.8 million a year earlier. It moved from a €23.7 million net loss in 2024 to €11.2 million of net income in 2025. Adjusted EBITDA reached €15.8 million.
No one should pretend those numbers make Trivago invincible. They do not. Its 2025 adjusted EBITDA margin was still thin, and a good quarter is not a permanent moat. But it is hard evidence that the turnaround has produced something more useful than a LinkedIn post about “transformation.”
Thomas’s management wager is especially interesting because he reportedly put a clean choice to Trivago’s supervisory board late last year: shrink the company to 200 people for efficiency, or keep roughly 600 people and use AI to create the impact of 6,000.
Most boards instinctively understand the first option. Cut headcount. Celebrate “operating leverage.” Put a slide in the investor deck. Everyone gets to call themselves disciplined.
The second option is messier. It requires leaders to make people more capable, change processes that have calcified over years, decide where humans still need to exercise judgment, and measure whether the output is genuinely better. You cannot delegate that to the chief technology officer and call it strategy.
AI is not the strategy. Better work is.
Thomas says he now spends more than half his time on AI transformation, compared with roughly a quarter a year ago. Good. That is where a CEO should be spending time if he believes the technology changes the economics and speed of the business.
But there is a trap here. Lots of executives hear that and conclude they need more AI meetings, more vendors and a company-wide mandate to “use AI.” That is how you end up paying for thousands of unused software seats while employees quietly carry on with the old process.
AI only creates value when it changes an economic or operational outcome that matters.
For Trivago, the sensible places are obvious: improving hotel discovery, producing better search and review summaries, improving conversion, making advertising spend work harder, and accelerating the teams building those systems. The company has said its AI-powered campaigns have operated in 30 countries, while branded traffic revenue has outpaced overall revenue growth. In its second-quarter results, it attributed revenue growth partly to stronger branded-channel traffic and improved booking conversion.
That is the standard. Not “we deployed AI.” The standard is: did more customers arrive directly, did more of them book, did we make a better product, and did the margin improve?
If an AI initiative cannot be connected to one of those answers, it is probably theatre.
I have seen this in businesses of every size. Teams get hypnotised by tools and forget the job. They start with the software rather than the bottleneck. Then they wonder why the promised productivity boom looks suspiciously like a few clever people making better PowerPoint slides.
Start with the expensive, slow or error-prone work. Find the handoffs where customers wait. Find the decisions that competent people repeat 50 times a day. Find the marketing dollars that go out before anyone can properly explain their return. Then decide whether AI can eliminate steps, improve judgment, or create more shots on goal.
That is management. The tech is merely the new power tool.
Trivago’s real advantage may be its size
The fashionable view is that AI will help the giants because they have the best engineers, the deepest data pools and more money than a pub full of mining executives on bonus day.
There is truth in that. Big companies can fund enormous infrastructure costs and recruit serious technical talent.
But they also have committees, legacy systems, turf wars, risk teams and layers of managers defending work that no longer needs doing. A company with 600 people can move faster than one with 60,000, provided its leaders are willing to make decisions.
Thomas’s argument is that medium-sized firms with a technology DNA sit in a sweet spot: enough scale and data to matter, but fewer bureaucratic layers and political obstacles to adoption. That rings true to me.
Small companies have urgency but often lack distribution, process and enough capital to sustain experiments. Big companies have assets but can take a year to agree on the font in the AI policy document. The middle can be dangerous if it is focused.
That does not mean every mid-sized company should mimic Trivago. Travel search is data-rich, digital, high-volume and measurable. An industrial business, professional-services firm or restaurant group will have different use cases and a different pace.
The transferable lesson is not “buy the same AI tools as Trivago.” It is “make the CEO own the operating redesign.”
A chief executive who hands AI to an innovation team is usually saying, without saying it, that it is not yet core to the business. A chief executive who spends half his time on it is telling the organisation that the work itself is being rebuilt.
Employees notice the difference immediately.
The overlooked risk: pretending 6,000-person output means 6,000-person wisdom
Here is the bit the AI evangelists tend to skip because it ruins the party.
More output is not automatically more value.
If you let AI produce ten times as much mediocre marketing, weak code, generic customer service or half-baked analysis, you have not built a superhuman company. You have built a faster factory for rubbish.
The 600-to-6,000 ambition only works if Trivago preserves the parts of the business that cannot be automated cheaply: taste, commercial judgment, customer empathy, accountability and the ability to recognise when the machine is confidently wrong.
Travel is a perfect example. A system can summarise thousands of hotel reviews. But a business still needs to decide what information helps a family travelling with two kids, what makes a price comparison trustworthy, and what level of recommendation earns repeat use rather than a one-off click.
That is why I would not judge this strategy by headcount alone. Stable headcount can be sensible, but it is not a trophy. The important measurement is revenue and gross profit per employee, paired with customer outcomes: conversion, repeat use, complaint rates and trust.
Trivago has given itself a useful commercial test. It wants to reach a 10% adjusted EBITDA margin by 2028. That is far more meaningful than vague language about becoming “AI-native.” A margin target forces the leadership team to prove that speed and improved output convert into durable economics.
The same should apply to every operator reading this. If your AI strategy has no operational target, no owner and no date by which you will judge it, it is not a strategy. It is a hobby with a budget.
What this means for you
You do not need 600 staff or €168.4 million in quarterly revenue to apply this tomorrow. You need the guts to stop treating AI as an IT project.
First, pick three workflows that directly affect cash: lead generation, sales follow-up, customer support, product delivery, procurement, reporting, whatever actually moves your business. Do not start with a company-wide tool rollout. Start where delay, labour or poor decisions cost you money.
Second, appoint one accountable operator for each workflow. Not a committee. One person whose job is to improve a specific metric within 90 days. Give them permission to kill old steps, not merely add AI on top of them.
Third, measure the before-and-after numbers brutally: hours per task, conversion rate, response time, error rate, cost per acquisition, customer retention, margin. If the numbers do not improve, stop congratulating yourselves and change the approach.
Fourth, keep humans where judgment compounds. Let AI draft, sort, summarise, analyse and test. Make your best people decide what deserves the company’s reputation, capital and customer trust.
Finally, spend your own time on it. If you run a business and AI can reshape the economics of your work, you do not get to outsource curiosity. Thomas’s most useful move may not be a particular model or campaign. It may be the decision to spend more than half his time on the transformation rather than treating it as somebody else’s problem.
That is the uncomfortable bit. AI will not save poorly managed companies. It will simply make the well-run ones faster, sharper and much harder to catch.