Success stories are pleasant to read, but they teach badly. The winner always tells a beautiful legend: the garage, the sleepless nights, the faith in a dream. The losers tell no legends, they quietly close the company and go back to a salaried job. Yet they hold the most valuable information of all: where the cliff was, the place where the road simply ended. I have spent many years building systems for businesses, and I have seen plenty of projects die not from a lack of talent but from the same, almost comically repetitive causes. So instead of another hymn to unicorns, I suggest a walk through the cemetery. Oddly enough, a cemetery is a very instructive place: nobody lies there.
A cemetery where nobody lies
There is an old story about bombers in the Second World War. Engineers studied the planes that came back, looked at where the bullet holes were densest, and planned to reinforce exactly those spots. The mathematician Abraham Wald told them to do the opposite: reinforce the places where the returning planes had NO holes. Because the planes hit there never came back. Startups work the same way. We study the survivors and copy their habits, while the real lessons are lying in the wreckage of those who never made it home. Below are ten pieces of wreckage that I find the most telling. You have probably heard the names, but the reasons are often retold wrong.
Ten pieces of wreckage
1. Webvan. Online grocery delivery in the dot-com era. The idea was absolutely right, and dozens of companies live on it today. But Webvan built giant automated warehouses for demand that did not exist yet. It is like building an airport in a village and hoping the planes will show up on their own. The infrastructure outran the market, and the money ran out before the customers arrived.
2. Pets.com. The famous sock puppet in the ads, huge brand recognition, and pet food sold for less than it cost including delivery. Every sale made the loss bigger. This is my favourite example of the fact that marketing can only accelerate a car that has an engine. If the engine runs in reverse, marketing simply drives you to bankruptcy faster.
3. Juicero. An internet-connected juicer costing hundreds of dollars, with juice packs on subscription. Then journalists showed that you could squeeze the pack by hand with the same result. The engineering was gorgeous, the problem was missing. A classic case of a solution looking for a pain to stick to.
4. Theranos. A promise to run hundreds of tests from a drop of blood. The technology did not work, and the company kept selling the promise until it all ended in court. This is no longer about business, it is about what happens when the pitch lives separately from the product. On a smaller scale I see this disease all the time: the slides exist, the working code does not, and everyone pretends these are the same thing.
5. Quibi. Short series for smartphones, Hollywood stars, nearly two billion dollars of investment, and a shutdown about six months after launch. The money could not buy the main thing: an answer to why a person should pay for something YouTube gives them for free. A big budget does not cancel the need to test a hypothesis, it only makes the failure more expensive.
6. Google Glass (the consumer version). A technical miracle and a social disaster. People nearby disliked the idea of being filmed, and the owners disliked the way people looked at them. The product was tested in the lab but not in a cafe. Later the glasses found a home in factories, where nobody is embarrassed, and that is a separate lesson about choosing your market.
7. Segway. We were told it would change the architecture of cities. It did not, because cities, sidewalks and laws turned out to be stronger than a scooter. Expectations were so inflated that even a decent niche product looked like a failure. Hype around a launch works like a loan: later you have to pay it back with interest.
8. Vine. Six-second videos, a brilliant format, an enormous audience. The platform gave creators no way to earn, the creators left for places that pay, and the viewers followed them. A community does not belong to you, you only rent it for as long as it gets more from you than it gives.
9. Color. An app for sharing photos with people nearby, with huge investment before anyone had used it. The product was built in isolation, users opened it, understood nothing and closed it. Big money at the start often hurts: it lets you avoid meeting reality for far too long.
10. Jawbone. Stylish speakers and fitness bands, a strong brand, but manufacturing with quality problems, endless legal wars with competitors and an attempt to fight on every front at once. The company spread itself thin and burned out. Focus in hardware costs more than in software, because every mistake is already cast in plastic.
What these wrecks have in common
Put these ten stories side by side and you see that technology was almost never the main problem. Webvan knew how to build warehouses, Juicero knew how to build mechanics, Google knows how to do everything. They fell for three reasons that repeat themselves, boringly, decade after decade. First: a product the market does not need in the form it was built. Second: the money ran out before the model started earning, because it was spent on scale rather than on validation. Third: team and management, when nobody could stop and say "wait, we are going the wrong way". None of these causes is cured by code. But all three are cured by the discipline of testing, and that is where code can help a lot.
Failure is almost never sudden. It stands in the corner of the office for years, and everyone walks around it because it is awkward to ask whose it is.
There is one more amusing observation. Almost all of these companies had too much money and too few questions. A poor startup gets feedback from reality for free and immediately: there is nothing to pay with, so go find a customer right now. A rich startup gets feedback from reality with a delay, but all at once. I am not calling for poverty, I am calling for money that buys knowledge rather than scenery.
How not to become example number eleven
When a client comes to me with an idea, the first question is always the same: what is the most dangerous assumption in this idea? Not the most interesting or the most beautiful one, but the one that brings everything down if it turns out to be false. For Juicero it was "people will pay for a machine they can do without". For Quibi it was "people need paid short-form content". Both could have been tested in weeks and for a small fraction of the money that was spent. After that it is simple: build the smallest thing that answers the question and show it to real people. If the answer is no, we saved years. If it is yes, we already have a first version and first users.
That is exactly why we at Ingello hold on so tightly to fast prototypes. With Fractal we describe a system instead of typing it by hand, and we get a working prototype in days rather than quarters. It is neither magic nor saving for the sake of saving, it is a way to let reality answer you before the money runs out. Warehouses, mass marketing and Hollywood stars can wait until the hypothesis is confirmed.
What to do tomorrow: write down, on a single sheet, the three assumptions your project stands on, and honestly mark which of them you have tested on real people and which ones simply seem obvious to you. The obvious is usually exactly where the shell hits. Reinforce that spot.
