The Start-up

The word itself is older than the meaning we now attach to it. People described new ventures as start-ups through much of the twentieth century, but the term meant roughly what it sounded like: a company that was just starting up. It carried no special theory of how such a company should behave, no built-in expectation of explosive growth or spectacular failure. A start-up was simply a business in its first stretch of life.

What changed was the arrival of a specific engine behind certain new companies. From the 1970s onward, and with real force through the 1990s, the venture capital industry built around Silicon Valley developed a distinct kind of bet: put money into a young company not to earn steady dividends, but in the hope that a small number of these companies would grow enormous and return many times the original investment. That funding model quietly redefined what the companies taking the money were supposed to do. They were not meant to be stable and profitable soon. They were meant to grow fast, find something huge, or fail trying.

Return to Blank's definition and sit with the verb: a start-up searches for its business model. This sounds like semantics until you notice how completely it changes what the daily work is supposed to be. In a company that already has a model, good management means execution — hitting targets, refining a known process, doing the thing reliably and a little better each quarter. In a start-up, the model itself is the unknown, so the work is not execution but discovery, and discovery looks like a lot of educated guessing followed by fast checking.

This is why the healthiest start-ups behave less like small businesses and more like laboratories. The founders hold a set of assumptions — about who the customer is, what problem is worth paying to solve, how the product should reach people, what they will pay. Every one of those assumptions is really a guess. The job is to test them cheaply and quickly, keep the guesses that survive contact with reality, and abandon the ones that do not. When founders talk about pivoting, this is what they mean: not flakiness, but the deliberate junking of a guess that failed the test.

Once the start-up is understood as a search rather than an operation, its notorious failure rate stops looking like a scandal and starts looking like arithmetic. The figures vary by how you count, but the shape is consistent: something in the range of two-thirds to three-quarters of venture-backed start-ups never return meaningful money to their investors, and the great majority of all new companies of this ambitious kind do not survive to become the self-sustaining machines they set out to build. This is not a sign that the people involved are unusually foolish. It is the expected result of the thing they are doing.

A search for something that may not exist has a built-in mortality rate, and no amount of talent removes it entirely. If a start-up is a bet that a repeatable, scalable model is out there waiting to be found, then a lot of those bets are simply wrong — the model is too small, the customers will not pay enough, the timing is off by five years, a bigger company gets there first. Failure, in this frame, is not the malfunction of the method. It is the cost of the method, priced in from the start. The venture funding that shaped the category assumes exactly this: it needs most of its companies to die so that the rare survivor can be enormous.

Come back to the two companies on their corners. The baker and the founder in the co-working space still look alike from the street, still work the same long hours, still hope the customers come. But they are doing opposite jobs. One is executing a model that already works, hoping to keep it running. The other is spending money and years searching for a model that may not exist, on the wager that finding it would be worth all the searches that come up empty.