The startup lifecycle

A startup is not in one mode forever. Early on, the job is to guess and learn. Later, the job is to keep up with demand. Treating those as the same job — hiring, ads, and process while you are still guessing — burns the runway you needed for search.

The path is messy, but the stages are useful. They tell you what "good work" looks like this quarter.

The big idea

Think of finding a restaurant location versus running a chain. Search is walking streets, counting foot traffic, and changing the menu. Scale is hiring cooks and signing leases. Doing the second job during the first one is how you go broke with a beautiful kitchen and no customers.

The analogy breaks because startups can "move the restaurant" — a pivot changes the customer or the job without throwing away every lesson.

Walking through it

1. Idea → early product. You are guessing. Build the smallest thing that tests the riskiest assumption. See startup ideas.
2. Search for fit. Most energy goes to talking to users and iterating. Hiring, PR, and "process" are usually early.
3. Trough of sorrow. Metrics are flat or noisy. Founders doubt the idea. This is where most companies die — or pivot.
4. Fit inflection. Retention stabilizes, organic demand appears, sales stop feeling like begging. The question flips from will anyone want this? to can we keep up?
5. Scaling. Hire, systematize go-to-market, harden infrastructure, watch unit economics. Mistakes here are expensive. Premature scaling is a top killer — see why startups fail.
6. Maturity. Growth slows toward market rate. The company looks more like a normal business: margins, efficiency, maybe an IPO or acquisition.

The job for engineers flips at the inflection. Pre-fit, the highest-leverage work is often shipping experiments, instrumenting retention, sitting on customer calls, and deleting features that do not move the needle. Over-building for 10× load you do not have yet starves learning speed. After fit, reliability, cost per customer, and hiring leverage start to matter.

Common mistakes

• Running a scale playbook during search. Departments, brand campaigns, and a sales army before repeatable demand.
• Treating the trough as proof the idea is dead. Flat noisy metrics are normal in search. Silence from users is not.
• Staying in search after pull is obvious. The failure then is capacity, not curiosity.
• A six-month platform rewrite pre-fit. Ask what learning it unlocks that a two-week hack would not.

Check yourself

A five-person team has polite pilots, no retention curve, and a plan to hire eight sellers next month. Which stage are they acting like, and which are they in?

They are acting like scale. They are still in search.

When does "do things that don't scale" stop being the default?

After fit, when the problem is keeping up with demand, not finding it.

Key takeaways

• Search and scale are different jobs. Confusing them is a common way to die.
• The trough of sorrow is where most companies quit or pivot. Noisy flat metrics are not the same as no demand.
• Fit flips the question from persuasion to capacity.
• Engineering should optimize for learning speed before fit, and for leverage and margin after.

Further learning

A practical early-company checklist:

Sam Altman: Startup PlaybookTo have a successful startup, you need: a great idea, a great market, a great team, a great product, and great execution.Sam Altman

Next, the trade you make by joining one of these companies: risk and equity.