I split my time between Menlo Park and Milan. Two cities, two time zones, two completely different relationships with failure.
In California, when someone tells you their startup didn't work out, the follow-up question is "what are you building next?" In Milan, it's closer to "what happened?" — said with the same tone you'd use at a funeral. Same word, different planets.
This isn't a cultural observation for LinkedIn. It's an operating system problem. And it's costing European startups years.
Let me be specific about what I mean, because the "Silicon Valley celebrates failure" narrative is mostly wrong. Nobody in Menlo Park is throwing parties because their company went under. What Silicon Valley actually does is something far more practical: it kills ideas early and cheap. The celebration, if you want to call it that, is for the speed of the kill — not the death itself.
There's a man named Alberto Savoia who understood this better than anyone. He was Google's first Engineering Director. He spent years watching products fail — not because they were badly engineered, but because they were the wrong product. His conclusion was blunt: most new products fail in the market, even if the execution is flawless. He called it the Law of Market Failure. And then he did something useful about it — he built a discipline around testing whether something should exist before anyone writes a line of code.
He called it pretotyping. Not prototyping. Pretotyping. As in: pretend the thing exists, see if anyone actually wants it.
I've spent the last few years bringing AI infrastructure into European financial institutions. I work with banks, wealth managers, asset managers — the kind of organizations where a product decision isn't a pivot, it's a three-year commitment with regulatory consequences. And the pattern I see, over and over, is the same.
European fintech founders build first. They build beautifully. They build with precision. And then they discover — 18 months and a few million euros later — that nobody wanted what they built.
I've watched a team in Germany spend two years building an AI-powered compliance tool. Perfect architecture, clean code, the works. They launched to silence. Not because the tech was bad. Because the compliance officers they were targeting didn't think they had a compliance problem. The founders had assumed the need. They never tested it.
Meanwhile, I've seen teams in the Valley validate the same kind of idea in two weeks. Not with a product. With a landing page, a fake button, and a credit card form. Ugly, fast, and brutally informative.
The Germans built a cathedral. The Californians built a cardboard box. The cardboard box told them more.
The XYZ Hypothesis
Savoia's framework gives this instinct a structure. The core idea is the XYZ Hypothesis: "At least X% of Y will Z." It sounds simple because it is. But try to fill it in for your own product and you'll see where the discipline bites.
"At least 5% of Italian private bankers who see our demo will request a pilot within two weeks."
That's a testable statement. You can run it in a week. You don't need a product, you need a presentation and a list of phone numbers. If 5% don't convert, you don't have a product problem — you have a premise problem. Go back to the drawing board before you touch a single database schema.
What Savoia understood, and what most European founders don't, is the difference between opinions and data. Everyone will tell you your idea is interesting. Your friends will say it's great. Your investors will nod. Even your target customers will say "yeah, I'd probably use that."
None of this is data. It's politeness.
Data is when someone pulls out their credit card. Data is when someone signs a letter of intent. Data is when someone changes their calendar to show up for your pilot. Skin in the game. Everything else is noise.
The Italian Trap
I want to talk about Italy specifically, because I live here and I care about this.
Italy has extraordinary talent. The engineering coming out of Politecnico di Milano, the design thinking, the artigiano mentality — this obsessive craftsmanship that makes Italian products beautiful. In execution, it's a superpower. An Italian engineer will build you something elegant where an American engineer will duct-tape something functional.
But in market validation, the artigiano mentality is a trap.
Because the craftsman falls in love with the object. The finish. The detail. The elegance of the solution. And falling in love with the solution before you've confirmed the problem is how you burn two years of runway on something nobody asked for.
I see it constantly in Italian fintech. Founders who can explain their architecture for forty-five minutes but can't tell you, with data, how many customers would pay for it. They've built the thing. They haven't tested the thesis.
The Valley founder does it backwards. Thesis first. Ugly test. Data. Then, only then, build.
It's not that Americans are smarter. They're not. It's that the ecosystem punishes late failure so brutally — through burn rate, through competition speed, through investor impatience — that early validation became a survival skill. Europe doesn't have the same selection pressure. So the muscle never developed.
AI Changes Both Sides
Here's where it gets interesting. AI is changing both sides of this equation, and not in the way most people think.
On one hand, AI makes pretotyping absurdly easy. You can now fake sophisticated product capabilities with a well-crafted prompt and an API call. You want to test whether wealth managers would use an AI assistant that summarizes client portfolios before meetings? You don't need to build the assistant. You need ChatGPT, a spreadsheet of sample data, and a PDF template. Run the test. See if anyone cares. The cost of faking a product has collapsed to nearly zero.
On the other hand — and this is the trap — AI makes it dangerously easy to fall in love with the technology. Because the demos are spectacular. You wire up a language model to your data and it feels like you've built something. The output looks real. It sounds intelligent. Your board is impressed. Your team is excited.
But "impressive demo" is not "market demand." I've sat in rooms where everyone was amazed by what the AI could do, and nobody had asked whether any customer would pay for it. The technology seduces you into skipping the validation step. It makes you feel like you're further along than you are. This is exactly the pilot trap I see in banking: impressive sandboxes, zero production deployments.
This is Savoia's Law of Market Failure wearing a new outfit. The failure mode hasn't changed — building the wrong thing. The speed at which you can build the wrong thing has just increased dramatically.
The founders I respect most right now, in both ecosystems, are the ones who use AI to validate faster, not to build faster. They're running ten pretotype experiments in the time it used to take to run one. They're using language models to simulate product experiences, gauge reactions, and collect real behavioral data before committing to architecture decisions.
They treat AI as a validation tool, not a construction tool. At least not yet. Construction comes after you know you're building the right thing.
The Business Plan Is Dead
I'll close with something that might be unpopular in European startup circles.
The business plan is dead. Not dying — dead. The 40-page document with market sizing and five-year projections that Italian and European founders still spend weeks writing? It's a fiction. It tells you what the founder hopes will happen, not what will happen. No business plan survives first contact with a real customer.
What should replace it is a one-page document with three things: a clear XYZ Hypothesis, a description of how you'll test it this week, and a definition of what "failure" looks like so you can't move the goalposts when the data comes back ugly.
That's it. That's the whole discipline. Savoia spent a career at Google refining it, and it fits on a napkin.
The hard part isn't understanding it. The hard part is having the honesty to let the data kill your idea before you've fallen in love with it.
Europe has the talent, the capital, and increasingly the infrastructure to compete with Silicon Valley. What it doesn't have — not yet — is the cultural permission to kill ideas fast. Menlo Park misreads this as Europe being slow, but the real issue is deeper than speed. To treat a failed experiment as Tuesday, not as a tragedy.
Until that changes, we'll keep building beautiful products that nobody wanted.
And that's the most expensive kind of failure there is.
