Okay great, you launched an API. Users are starting to consume your API. But how do you measure its success? Although the metrics described below are universal in determining how well your API is doing, the ones you decide on should ultimately be driven by your company’s core competencies and objectives.
Performance has to do with the speed and quality of your API. On average, how long does it take for your API to respond to calls taking into consideration the size of the request and response? How does that response time affect the user experience? Your response time will vary depending on what your are delivering. In the case of a mobile experience, speed is critical, and so responses should be delivered in under half a second. When running sizeable operational reports, the response time can be anywhere from two to four times longer. When the API response time is more than 10 seconds, you should consider providing the user with a progress bar. This feature can help the user plan his or her time accordingly and rest reassured that the system has not crashed.
In regards to quality, what percentage of your calls return error messages? If the API request results in an error, is your documentation obsolete or misleading? The ultimate goal of an API is to deliver the right information in a simple, scalable way. Without proper delivery, the API is not achieving its objective and therefore not much of a success. If you don’t have one already, consider launching an API status page to inform users of the API’s availability and why it is down in the (hopefully rare) event that it is.
Quantity of Requests
The quantity of requests is an important metric to consider. A large, growing number of requests is usually a sign of success. Having said this, an increase in the number of requests could also mean that your API is inefficient; if users have to send multiple requests for a single query, then this can also result in a surge of traffic. So it is important to understand why you are experiencing heavy call volume.
Your API will undoubtedly face a handful of operational problems. The speed at which you handle these issues can impact many of the factors described above including API performance and efficiency and the quantity and quality of call requests and responses. By resolving issues quickly, you can help shorten the downtime and encourage developers to continue using it. Although issues will always arise, it’s important to note that a well-designed code architecture can help prevent many of these problems.
In addition to the list above, your company should also consider tracking the following metrics:
Calls per Method
Availability of Each API Endpoint
Type of Export Format (Json, XML)
Channel Used (web, mobile, etc.)
Breakdown of Programming Language
Errors per API Endpoint
Percentage of Error
Breakdown of Response Percentage
We hope this post helps clarify some of the essential elements of a successful API. Have questions? Need advice? Drop us a line or contact firstname.lastname@example.org!
SwiftIQ provides technology to turn your company’s data into a platform to create intelligent, data-driven applications and predictive insights.