These recent years, as the internet users became saturated and increase more and more slowly, the network traffic is much more costly than ever before, only making a user-friendly internet product satisfying the market demand is not enough to pursue success. We need to make sure every click counts.
Here we have AARRR startup metrics, which are divided up into five sections – acquisition, activation, retention, revenue, referral, to analyze an internet product. However, actually, things do not always happen in the exact order, the last three sections can be processed in different priorities base on the purpose of your product in different stage. For instance, if your product is still in the MVP stage, the level of retention is your first priority. Because you need to make sure whether customers buy your ideas. Until you get adequate data convincing your thought, try not to pay much effort on other sections.
To analyze the product, it would be much easier if you have a reliable data platform for you to trace the performance of all the traffic flows. Many analysts tend to assess the product in noddle, driving them to crazy with nothing convictive found. As a debatable conclusion will not lead to a helpful insight, making sure you can trace the detail of your funnel from the different channels will save you some unnecessary arguments.
If it was a web product, using the designed trace code attaching the link will be a good idea. Instead of building your own data platform, you can make use of some third-party platform services like google analysis to trace your link.
If it was an app, using codes to mark your Android install package for different traffic flows. When it comes to the iOS app, which people can only download from the APP store, we need users to leave something unique for us to match the traffic flow and then trace. For example, requiring users to register to get coins for reward, we attach the register information to the specific traffic flow.
In order to make the most out of our data, we should try to make enough tags for our users for Muti-dimension analysis. I will carry it out in my next article.