This article takes "how to evaluate the effect of strategic products" as the object of discussion, and believes that it can be evaluated at different stages, combining user feedback, embedding, AB testing and other methods. New product managers text message service should remember to shift from "qualitative thinking" to "quantitative thinking". Especially for strategic products, real-time data monitoring should be carried out throughout the product life cycle. So how to predict and evaluate the effect of a new strategy? The following are some of my thoughts and insights, hoping to help everyone. 1. How the strategy was born Before talking about strategy metrics, we need to understand how strategies are born? A strategy has the following four elements: Problem to be solved:
the most essential purpose of our strategy Input: that is, a series of factors that affect the formulation of strategies obtained by considering the feelings and demands of different users in different scenarios Computational logic: that is, how do we take text message service advantage of these factors and assign them appropriate weights and calculation rules Output: i.e. the specific solution to the problem The following can illustrate the specific application of the four elements with an example: Feed flow intelligent recommendation strategy 1. Problems to be solved: Provide more appropriate and interesting information recommendations for different users. 2. Enter: User characteristics, which may include the user's gender, age, searched content, viewing time, and favorite information.
Information labels, such as the keywords included in the information, the matching situation of current popularity (such as celebrity marriage and text message service divorce always lead the information hotspot). 3. Calculation logic: By assigning different weights to the different inputs, a multi-factor formula is set to finally obtain the quantitative index of user favorability. 4. Output The significance of establishing metrics By understanding the four elements, we can abstract the meaning of establishing measurement indicators into the following points: Problems to be solved: Determine whether the focus and positioning of the “problems to be solved” are accurate Input: Verify that the multi-factors of the input are sufficient and appropriate Computational logic: check whether the assignment weight is appropriate Output: Determine whether this strategy effectively solves the problem,