What should founders do in the first 90 days of a startup?

In the first 90 days, founders should do almost nothing except talk to users and build. The single biggest mistake is treating your fragile early startup by the standards of an established company — it will never survive that comparison, and it doesn't need to. Your only job is to find out whether you're building something people actually want, and to survive long enough to find out.

Accept fragility as the baseline, not a warning sign

Every early-stage startup is fragile. That's not a red flag — it's the normal condition of anything new. The danger Paul Graham identifies in his essay on doing things that don't scale isn't external critics or skeptical investors; it's founders dismissing their own startups before giving them a real chance. Reporters, forum commenters, and even some investors will judge your early product by the standards of something mature, and that judgment is almost always wrong and almost always irrelevant.

What matters in the first 90 days is that you don't make the same mistake yourself. The fragility of your product, your team, and your user base isn't evidence that the startup won't work. It's simply evidence that you're early. The correct response to fragility is not to panic or pivot wildly — it's to apply direct, personal pressure to the problem. Get close to your first users. Do the things that wouldn't work at scale. That's not a workaround; that's the job.

This means resisting the urge to professionalize too early. Founders who spend their first 90 days writing internal processes, building org charts, or optimizing for investor optics are solving problems they don't yet have. The only existential question in this window is whether real people will use what you're building.

Talk to users in person — more than feels necessary

The most leveraged thing you can do in the first 90 days is meet your users face to face. Not survey them, not read analytics about them — actually sit with them, watch them struggle, and listen to what they say and what they don't say. This feels inefficient when you think about scale, but that's exactly why it works: you can absorb more signal in one hour of direct observation than in weeks of remote data collection.

The Airbnb example Paul Graham returns to repeatedly in his writing is instructive here. Their early survival came down to roughly a month of founders going out and personally engaging with hosts and guests. That direct contact didn't just generate feedback — it built the kind of trust and iteration loop that remote interaction can't replicate. The key insight is that the quality of your early user relationships determines the quality of the product you'll build. Generic feedback produces generic products.

In practice, this means targeting your first ten to thirty users with obsessive care. Find them through any channel that works — direct outreach, communities, your own network — and give them a level of personal attention you will never be able to give at scale. This isn't a growth strategy; it's a learning strategy. You're trying to figure out what problem you're actually solving before you scale the wrong thing.

Don't fundraise in the first 90 days unless you have to

Fundraising is one of the most time-consuming things a founder can do, and it grinds nearly everything else to a halt. Paul Graham notes that for a small founding team, a fundraising process can easily consume six weeks — six weeks when you could be building, talking to users, and iterating. Unless you're running out of runway immediately, the first 90 days should be about making your startup into something worth funding, not chasing checks.

The practical implication is that you should reach a clear product milestone or user traction signal before initiating fundraising conversations. Investors can tell the difference between founders who are raising because they need validation and founders who are raising because they've found something real and need fuel. The latter have significantly more leverage — and as Graham observes in his writing on investment trends, founders who demonstrate genuine product-market traction increasingly have the upper hand in those conversations.

If you do need to raise early, keep it short and focused. Talk to investors in parallel, not sequentially, so you compress the timeline. Don't let a single conversation drag on for weeks. Your goal is to close a round in days, not months, so you can return to the actual work of building.

Be honest about what you don't know — especially with yourself

One of the subtler lessons from Paul Graham's writing on early-stage startups is that the founders who make the best decisions in their first 90 days are the ones who accurately assess their own uncertainty. They know what they know and what they're guessing. They can articulate their competitors' real strengths, not just dismiss them. They hold their hypotheses loosely while committing hard to execution.

This matters practically because the first 90 days is fundamentally a hypothesis-testing period. You have a theory about who your user is, what problem they have, and why your solution is better than the alternatives. Almost none of those hypotheses will survive contact with reality unchanged. Founders who are too attached to their original assumptions waste those 90 days defending a vision rather than updating it.

The antidote is to build explicit feedback loops into your weekly rhythm. After every user conversation, write down what you learned that contradicts your assumptions. Track which features users actually use versus the ones you thought they'd care about. Make it structurally easy to be wrong, because being wrong early and cheaply is the entire point. The founders who adapt fastest in this window are almost always the ones who are still standing a year later.

Measure one thing that actually tells you if you're working

In the first 90 days, most metrics are noise. Traffic, sign-ups, social followers — these numbers feel like progress but rarely tell you whether you're building something people need. The one measurement that matters is some version of retention or repeated engagement: do the people who tried your product come back? Do they tell someone else about it without being asked? Do they complain loudly when it breaks, which is its own form of engagement signal?

Choose a single metric that represents real value delivered to a real user, and track it weekly. If it's moving up, you're learning something worth reinforcing. If it's flat or declining, that's the most important signal you can have — and it's far better to discover it in week three than in month twelve. The metric doesn't need to be large; it needs to be honest.

This focus on a single metric also helps you prioritize everything else. When deciding whether to build a feature, add an integration, or spend a day doing outreach, the filter is simple: does this move the metric? If not, it can wait. The first 90 days is too short and too critical to spend on anything that doesn't directly connect to whether your core product is working.

“Almost all startups are fragile initially… the big danger is that you'll dismiss your startup yourself.”

— Paul Graham, source

The one thing to do

Spend the first 90 days talking to users in person and tracking one honest retention metric — everything else can wait.

Frequently asked questions

Should founders be focused on growth or learning in the first 90 days?

Learning. Premature growth locks in assumptions you haven't validated yet. Focus on understanding your users deeply enough that growth becomes inevitable, not on manufacturing it artificially.

How many users do you actually need in the first 90 days?

Fewer than you think. Ten deeply engaged users who genuinely rely on your product teach you more than a thousand passive sign-ups. Depth of relationship matters more than breadth at this stage.

Is it a bad sign if the product still feels rough at day 90?

Not necessarily. A rough product that real users find indispensable is far more valuable than a polished product nobody needs. Roughness is fine; lack of genuine user engagement is not.

When should fundraising enter the picture for early founders?

After you have something concrete to show — either meaningful retention, a sharp articulation of a real problem, or ideally both. Fundraising before that usually costs more time and equity than waiting a few more weeks would.

Sources

More playbook answers · Growth Prophet home