Hopefully, a lot of traffic and business will come through the API. As a result, you’ll need metrics for a variety of reasons, including operational reasons as well as for measuring the success of the API and gauging how it is being used.
Let’s face it: It’s hard to improve what you don’t measure, so we suggest baking your business model into the API. For example, if your company sells ads, you need to know how the API affects your ad-supported business. If your company sells cars, you need to know the API supports your company’s ability to sell cars effectively.
Different types of APIs need to measure different things. Some of what we discuss in this chapter may not apply to your API (for example, content-focused APIs need different metrics than transactional APIs). To add an extra twist, what you measure will change over time. The point is to establish some metrics as a starting point, then see what others you need.
The bottom line is that analytics provide you with information to make future decisions. That information might mean a variety of different things. It might indicate that traffic is up because the API is succeeding, but a surge of traffic could mean that the API is being used inefficiently, resulting in inflated traffic.
The more you know, the better you’ll be able to see and demonstrate what is going on. You’ll need metrics to help you decide which features to add and how to evolve the API. And sometimes you’ll have a hunch ...