What makes early-stage startups fail, and how to avoid it

Executive overview

Most early-stage startups fail for one reason: they don't talk to users, so they never find product-market fit. Gustaf Alströmer has worked with over 600 YC companies and sees this pattern repeat constantly — founders confuse external validation (investor interest, press) with actual customer demand.

The fix isn't a framework. It's a discipline: get in front of real users before building, watch them work, and move fast based on what you learn.

The hardest thing for technical founders isn't building — it's accepting that talking to customers is the actual job.

Why startups fail

• Not talking to users is the root cause of almost every failure
• Technical founders default to building; they treat customer conversations as optional
• Founders confuse YC acceptance, investor interest, or press coverage with product-market fit
• YC reminds every new batch on day one: almost none of you have product-market fit
• The fear of rejection stops founders from doing outreach — but most non-users are indifferent, not hostile
• Indifference is the most common outcome; it doesn't close the door, it just means they're not ready yet
• Non-technical founding teams are a strong failure signal; YC rarely accepts teams that can't build their first prototype

What YC office hours actually do

• Individual office hours start with one question: "What's holding you back from moving faster?"
• The goal isn't updates or strategy — it's surfacing blockers the founder hasn't named yet
• Group office hours create accountability: what were your goals, did you hit them, what got in the way?
• Founders are competitive; they don't want to come back with nothing learned
• Starting a company is profoundly lonely — employees can't absorb founder anxiety, investors aren't available
• Group settings let founders see that every company is broken in some way; normalises the experience
• The most productive YC periods happen when founders meet in person, no laptops, full attention

How to talk to customers

• Reach out volume matters: expect to contact 10 people to find 1 early adopter
• Early adopters are roughly 10% of any market; the other 90% aren't incentivised to try new things
• The outreach itself is sales — people you talk to are often your first customers
• Don't ask "do you have this problem?" — watch them work instead
• Screen sharing or walking through a daily workflow reveals pain that users can't articulate
• Example: rent an EV and use every non-Tesla charger; the product quality speaks for itself
• Target 25–50 conversations minimum; high-performing founders often reach 100–200 before raising

On technical co-founders

• Engineering is the hard part — valuing it at 10% equity signals you don't understand this
• A contracting firm can't iterate fast enough; someone has to own the product decisions
• Technical co-founders also want business partners — rejection rates are high but ask your best people
• Non-technical founders can succeed if the engineering team feels like a founding team, not contractors
• Learning to code yourself is a viable path; you don't need to be a great engineer, just good enough to understand what engineering enables
• YC has a co-founder matching platform for founders without a technical partner

Attributes of successful founders

• Determination: won't stop when things are hard; the drive is internal, not for investors
• Technical competence: more technical founders succeed at a meaningfully higher rate
• Move fast without waiting for permission — from investors, from YC, from anyone
• Excellent communication and storytelling: motivates teams, enables fundraising, attracts talent
• Projecting confidence while maintaining genuine humility — both are required, they're not in conflict
• Execution focus over strategy: strategy assumes you can do multiple things; early startups can only do one

What YC can and can't predict

• YC is good at recognising failure patterns, not at identifying the next Airbnb
• The clearest positive signal during a batch: each office hour brings genuinely new progress, not the same problems
• Weekly or bi-weekly progress is a stronger indicator than market size or investor interest
• Outlier companies often have ideas that seem illogical — investors passed on Airbnb repeatedly
• YC's job is to prevent smart teams from making common mistakes, not to pick winners

What made Airbnb's culture exceptional

• Founders hired people who were genuinely excited to be there, not just impressive on paper
• Many early PMs were former founders — shaped how decisions were made and risks taken
• Culture interviews were used to screen for values alignment, not just skills
• The company had such obvious product-market fit early that the team could take risks freely
• The alumni bond resembles close friendships more than former colleagues

Climate tech: the shift and the opportunity

• The decision to decarbonise has already been made by governments and large corporations
• This transition involves trillions of dollars moving from carbon-emitting to clean alternatives — larger than the software industry
• The IRA in the US and equivalent European responses are now pulling private capital into the space
• Fortune 100 companies now show up as paying customers for decarbonisation software, not just explorers
• Corporations are afraid of being "the Toyota" — the last major player that didn't adapt
• Software skills transfer directly; domain expertise in climate science is complementary, not required

Areas and companies worth watching

• Carbon accounting: Unravel Carbon (Asia), Carbon Chain (supply chain/shipping), Sinai (Bay Area)
• EV infrastructure: Enode (API layer for EV chargers and home energy systems), Static (EV charging network in India)
• Shipping: Seabound (onboard carbon capture for freight ships), Fleet Zero (electric ships)
• Aviation: Hard Airspace, Wright Electric (battery-electric commercial aircraft)
• Carbon removal: a parallel workstream alongside decarbonisation — likely necessary even if decarbonisation goes well
• YC's updated request for startups (written with David Rusenko) lists 25+ specific areas for founders without domain expertise