Business, explained

Why most startups fail

The common cause is not bad technology. It is building something people will use but not pay for, or pay for exactly once.

The signal that separates a business from a project is retention: customers who come back without being re-acquired. Everything else can be bought. Traffic can be bought, signups can be bought, even revenue can be bought if you are willing to lose money on each sale. Retention cannot, because it is a measurement of whether the product was worth returning to.

This is why the order of operations matters so much. Founders who spend on growth before retention is proven are buying a bigger leak: every marketing pound pushes users into a product they will abandon, and the churn scales with the spend. Founders who fix retention first spend less overall, because each acquired customer stays long enough to pay for the next one.

The failure sequence, in order

1. A real problem is identified, or assumed.
2. A product is built for it, usually well.
3. Early users try it, and some say encouraging things.
4. Usage decays quietly. Nobody complains; they simply stop.
5. The team reads flat growth as a marketing problem and raises to fix it.
6. Paid acquisition scales the leak until the money runs out.

Step four is where the business was decided. Steps five and six are the expensive part of discovering it.

Moats, in one paragraph

A moat is any reason a competitor with equal funding and equal talent still cannot take your customers. In practice there are about five.

Everything else — a better interface, a two-year head start, a clever feature, a talented team — is a lead, not a moat. Leads are valuable, and leads expire. The strategic question is always whether you are converting your lead into one of the five before it does.

Unit economics, without the spreadsheet

Take exactly one customer. Add what it cost to acquire them and what it costs to serve them for as long as they stay. Subtract that total from everything they pay you over the same period.

If the result is negative, scale makes things worse, not better — you are buying revenue at a loss and the loss grows with volume. Nearly every argument about growth strategy eventually reduces to this one calculation, and most of them would be shorter if it were done first.

The three numbers

• Acquisition cost. All sales and marketing spend in a period, divided by customers gained in that period. Include salaries; excluding them is the most common way this number gets flattered.
• Gross margin per customer. Revenue minus the cost of actually delivering the service — hosting, support, payment fees. Not net profit; delivery cost only.
• Lifetime. How long they stay, derived from churn. A 5% monthly churn rate implies an average life of about twenty months.

Multiply margin by lifetime, compare to acquisition cost. A ratio comfortably above one, with the payback happening in months rather than years, is a business. Below one is a subsidy.

Reading a job offer properly

• Base salary is the only number you can spend. Treat bonus, equity and profit share as conditional, because they are.
• Equity is meaningless without three figures: your share count, the strike price, and the total shares outstanding. A grant of "10,000 shares" tells you nothing until you know whether the company has one million or one hundred million.
• Ask what the last three people in this role went on to do. The answer, and the hesitation before it, is the most informative question in an interview.
• The growth rate of the team you join predicts your options better than the size of the company. New roles open where headcount is expanding.
• Vesting schedules and cliffs are negotiable more often than candidates assume, and are rarely asked about.