Thinking about risk

- When you're blitzscaling, it's really important to be a responsible blitzscaler. That's why we've given you a framework for evaluating risks. The key distinction is whether the risk is systemic or non-systemic. A systemic risk is something where the failure of that part of the system can bring down the entire system. If you're a company like Visa or MasterCard where you're actually operating the payments that undergird our entire economy, that's potentially a systemic risk. Non-systemic risk is far more typical, and it's actually something you can tolerate as a blitzscaler. If your product failure means that something goes wrong in a very localized area with a minimal impact, that's something that's acceptable. There's four ways that you can approach mitigating risk. The first is to take decisive action immediately. You reserve this for risks that are an existential threat to the company. For example, a number of years ago, an Airbnb host came home and discovered that the guest had trashed their home. This began to spiral out of control, and Airbnb's CEO, Brian Chesky had to step in and take decisive action guaranteeing that Airbnb would make good any damages that occurred to hosts. The second approach is to take short-term action to mitigate the immediate effects and defer a full solution until later. One example of this is PayPal and how it dealt with fraudulent transactions. It would've been very difficult to completely prevent fraudulent transactions from happening, but on the other hand, the problem was severe enough that it required some sort of response. So the decisions was reimburse the people that have been affected so that the short-term problem is taken care of, and then in the long run, fix the fraud problem before it bankrupted the company. The third approach is to note the problem now and then take action later. You use this for things where there is a problem and it will become big, but it doesn't require immediate action. So in the case of PayPal, there were a number of illegal transactions that might occur. People might use PayPal for some sort of crime. Well, was that an issue? Turns out the number of those transaction was vanishingly small. So it was sufficient to just say listen, we're going to note the problem, and we're going to fix it later on. The final approach to addressing a problem if it's non-systemic is to just let it burn. Blitzscaling is all about speed, and every time you take the time to try to solve a problem, you're slowing down. There are some problems that are small enough, it's worth it to let it burn so you can move faster.