Start with who . Tools are designed for a human to use them. Period. And this does not only apply to tools that humans apply to a problem directly. This also applies to tools that are designed to create other tools. As long as a human being needs to explain to another human being why he or she needs your tool, this tool is designed for a human being. I often hear from clients that innovation does not really happen in their industry because they are operating in a B2B versus a B2C environment. It does not matter. You are working and selling to human beings, so what you design has to appeal to them. Humans do not buy technologies; they buy tools they can justify and explain. If you help them with that by designing the arguments for the use case for your tool into the tool itself, you exponentially increase the likelihood of your tool being utilized.
Serve purpose first . Do not lose track of why you are designing it in the first place. If your purpose is to help people connect in more meaningful ways with friends, but then keep users on your platform by distracting them with games, news and videos of puppies, you are working against your purpose despite fulfilling a different goal. Some platforms make you addicted to their social feed, but it does not help you to connect in more meaningful ways with other humans. The bigger a company gets, the more they are in danger of losing focus as to why they exist. Ultimately, they start abusing their power over users until a big backlash hits them, or when another platform starts serving the original purpose in a better, more functional, or less addictive way.
Follow the user . When you notice that your users start using your tool for a different, sometimes unintended, use case, stay true to your purpose but don’t try to change your users. Your tool did not solve the problem you had in mind, but instead your tool helped solve another problem. Whatever your user thinks is the most useful case for your tool is automatically the most useful case. Resist the urge to explain to your users to change the way they use your tool. Stanford’s Andrew Ng explains in his lectures the difference between a retailer that opens an online store and an Internet retail company by pointing out that the latter uses all possible ways it can to follow the user. Rigorous A/B testing on the platform ensures that it constantly follows whatever users want to do. If their habits or short-term needs change so does the platform. Offline retailers who go online usually do not go through this same process, because they are used to a much slower adaptation process since they are used to dealing with physical environments that are much harder to change. Think of a major retail store with two locations and an A/B test of the entrance door. Even if location A of the two retail locations generates more revenue, you do not know if this is because of the difference in the design of the entrance or something else, unless every other factor between the two locations is exactly the same (which is impossible). The online retailer can theoretically test several entrance pages on thousands of users individually, take the best ones, turn them into variations of the most successful entrance, test them all out, and within days have a statistically significant decision about which entrance generates more revenue. Due to the experimental environment and massive data capabilities online, the Internet retailer will simply know more accurately and precisely what to offer to follow the user wherever they go.