As a product manager, focusing on data is an essential job. At work, there are often fluctuations in the indicators or data that we are concerned about, or when we find incomprehensible data when analyzing a function. At this time, we need to analyze the reasons for the fluctuations in the data, and I will also share my analysis ideas with you. . 1. Confirm the accuracy bulk sms service of the data source Many times the problem with the data is just the problem with the data. If we encounter a data fluctuation or anomaly and we cannot simply analyze the cause, we may have to think about whether there is a problem with the data source. The data source may be our business database, such as bulk sms service commodity prices, sales data; it may also come from user behavior, such as user clicks, browsing and other events,
These data sources are very likely to have problems, such as business database synchronization problems, Inaccurate collection and reporting of user behavior, etc. Sometimes we analyze from the product point of view, and in the end it is true that the problem of the bulk sms service data itself will do a lot of useless work. Therefore, if you encounter a problem that cannot be analyzed for a while, you must first consider whether there is a problem with the data source. At this time, you have to contact the company's data product manager, front-end and back-end behavior reporting parties, and big data research and development confirmation. There are several common factors that can lead to data source bulk sms service accuracy problems: The event reported by the front end does not match the definition of the product: for example, some pages may be preloaded.
From the perspective of the product, preloading is a technical processing method, not a user's behavior. However, the front-end often reports preloaded events indiscriminately, resulting in data that does not match expectations. Server refactoring/technical bulk sms service optimization: Server engineers will perform technical optimization from time to time. This part of the optimization products is often difficult to perceive, and at the same time, it is easy to cause data fluctuations. For example, there was a sudden increase in the amount of Push sending by our company. The positioning problem found that the technical team optimized the server that sends Push. After optimization, the target number bulk sms service of users remained unchanged, and the amount of sending per unit time (QPS) increased. During the period (such as 20:00~22:00), the push that has not been sent can now be sent in full. Big data statistical logic: