After the commercialization was involved, we put some high-quality charging templates in the algorithm to the top, and found that it was of great help to the revenue, and the effect was obvious. So we immediately made a manual operation of special database the top function, and a special person was responsible for it. After a period of trial operation, a lot of experience has been summed up to help us continuously improve the optimization algorithm. At the same time, it was found that some types have strong ability to leverage payment, and some types have weak ability to leverage payment. For example, the following is the ranking of the ability to leverage payment by category: In response to this situation, we will let the types with strong payment ability continue to drive revenue, and other types will drive template usage and downloads. Ultimately, we will achieve an optimal balance between revenue and template usage. 2. Conversion rate For special database ease of understanding, take Taobao, which is more familiar to everyone, as an example. Roughly divide its most important levels into: home page search result page product detail page order submission page payment page (the actual situation is much more complicated than this, but it is omitted for ease of understanding). At this time, we need to see what the conversion rate of "Home Search Results Page" is, let's say 95%; and then look at the conversion rate of "Search Results Page Product Details Page", let's say 80%; And look at the conversion rate of "product details page submit order page", let's say 20%; and look at the conversion rate of "submit order page payment special database page", let's say 60%. Then sort out the strategies that can improve the conversion rate at each level. Evaluate its cost and value, and prioritize low-cost, high-efficiency strategies. In 2014, when I was on Qunar.com, I did a similar project. At that time, after only one iteration, the down-conversion rate of the homepage was increased from 25% to 68%.