Data analysis of Internet finance platforms: These three models are enough

Since the operations department is the department in the Internet finance platform company that is responsible for performance and KPI, this means that operations personnel need to observe the data at all times and find abnormal situations in the company's operations based on the data.

If God blesses you with good luck and the business data growth is in line with expectations, then it would be great and you can eat chicken tonight; if there are disasters and the business data declines seriously and becomes abnormal, there is no need to panic. Learn to utilize and use the data, and through multi-dimensional comparative analysis of the data, find out the potential factors of data fluctuations, and think about the corresponding operational strategies and make adjustments quickly to ensure that the company's business returns to normal. As the saying goes, when the enemy comes, we will stop him; when the water comes, we will block it with earth.

The operational work of Internet finance products on the market is mainly divided into two types: one is the acquisition of funds (i.e. lenders or investors ), and the other is the acquisition of assets (i.e. borrowers). Typical examples of the former are securities trading, bank financial management, and P2P Internet finance products, while typical examples of the latter are cash loan and borrowing Internet finance products. The research object of this article is the former. Through three common data analysis models, this article explains the data analysis perspectives of Internet finance products obtained by the funding side. I hope it will be helpful to everyone.

Model 1, AARRR model , for the entire product level:

When analyzing from the perspective of daily product operations, the author applies the AARRR operation model that is currently popular on the market. This model can be applied to almost all Internet products, and you can also try to apply the analysis in your own vertical field.

Acquisition

Key data indicators: number of registered users, number of investment users, and average investment amount per person;

Analysis angle: It is necessary to compare the registration volume and registration quality of different channels . The registration volume depends on the number of registered users, and the registration quality depends on the number of invested users and the average investment amount per person (i.e. conversion rate ).

Increase activity

Key data indicators: Pageviews (PV), number of visitors (UV), number of IP addresses, daily active users of the APP, and number of daily APP launches;

Analysis angle: Different from social or information products, Internet finance products do not attach much importance to the activity indicator. Because from the most ideal situation, after investing, users do not need to open the APP to perform any operations, and then they can just wait for the principal and interest to be returned. This is also a reflection of the user's trust and confidence in the investment product. So when is the activity indicator worthy of operators' special attention? There are two situations: when major positive or negative news is released. There may be major positive factors such as platform financing and unprecedented marketing activities , or major negative factors such as the exposure of various negative news. At this time, operators need to pay attention to the fluctuations in activity and analyze whether the fluctuations are within the expected range.

Revenue

Key data indicators: number of investment users, transaction amount, average investment period, and time to full mark;

Analysis perspective: Revenue data corresponds to turnover on e-commerce platforms and to investment data on Internet finance platforms. A platform’s ability to attract money can be seen directly from the two main indicators: the number of investment users and the transaction volume. There are also two other auxiliary indicators: one is the investment period, and the other is the full-scale time. To borrow a phrase from Ma Huateng ’s father, “Finance is not about who runs faster, but who runs longer.” Therefore, the longer the investment period, it means that users trust the platform and are willing to invest for a long period of time, which also means that the platform’s operating conditions are better. The shorter the full-standard time, it means that assets are in short supply and it is relatively easy for the platform to obtain investment funds, which also means that the platform’s operating conditions are good.

Improve retention

Key data indicators: recharge amount, withdrawal amount, recharge and withdrawal rate, return amount, number of users with return, retention amount, fund retention rate, fund reinvestment rate, number of retained users, retained user ratio, churn amount, number of churned users, net fund inflow/outflow

Analysis angle: Retention is the top priority for Internet finance platforms. There are many data indicators involved, which may make people feel dazzled, but the core is to focus on two data, capital retention rate and capital reinvestment rate, because these two items are actually a combination of collection, withdrawal, recharge and investment. In the short term, the retention rate and reinvestment rate will fluctuate dramatically with short-term marketing activities; in the long term, the retention rate and reinvestment rate will steadily increase with the improvement of the platform's credit, interest rates, product experience and other aspects, and vice versa.

Self-propagation (Refer)

Key data indicators: number of invited users, investment amount of invited users, number of red envelope sharing, number of activity sharing

Analysis angle: Doing a good job of self-propagation can help the platform save a lot of channel costs and promotion costs, and pass this part of the costs on to the inviting users. The invitation function is almost one of the necessary functions of many Internet finance platforms. In addition, some large-scale activities and functions also have the value of sharing and dissemination. The most ruthless one I have ever seen is Ping An Bank. Almost all landing pages in its mobile app have sharing functions, reminding users to share and spread the word all the time. At the data level, just focus on the corresponding sharing data and conversion data, which is relatively simple.

Model 2, the proportion model, is for a certain functional level:

In addition to data analysis required for daily product operations, when new versions of Internet finance products are launched and new features such as VIP system, points system , task system, automatic bidding, and smart investment advisors are updated, operations also need to track and analyze data for specific features to understand user feedback on new features.

Some operators like to conduct follow-up surveys on users to understand their acceptance, but the author believes that only data is the most objective and real, and functions that can effectively solve user needs will definitely be well reflected at the data level. Here are two ways to distinguish:

1. Is the new feature popular?

Metric: Active ratio. That is: the number of active users using the new function / the number of active users in the same period.

2. Will users reuse the product?

Metric: Reuse ratio. That is: the number of users who continue to use the new feature on day N/the number of users who use the new feature on the first day.

(Part of the above content is quoted from the "Yilin Xiaoyu" column)

Model 3, the funnel model , is aimed at the operational activity level:

Activities are the daily work of operations personnel. If an activity is compared to a project, any complex project can be broken down through WBS: Work Breakdown Structure.

Similarly, even the most complex activity can be broken down into multiple steps for analysis. The commonly used analysis method here is to conduct a funnel model analysis to analyze the changing trend of the number of users from potential users to end users, so as to find the best optimization space. For example, the following example analyzes the changing trend from when a user enters an activity page to when they finally generate investment behavior.

Finally, we analyze this funnel to see whether the conversion rate of each link has met expectations. If not, you need to reflect on which parts of the activity process can be optimized, such as whether the copy is attractive enough, whether the steps are concise enough, and whether the experience is smooth enough.

Data can lie

I believe everyone has heard the saying that data can lie. In fact, the data itself will not lie, but the methods of collecting and analyzing data will lie. During the data analysis process, if we combine other dimensions for further analysis, we may sometimes come to completely different conclusions.

For example, a user operation team wants to conduct a user preference survey, so they call all the users who use their product every day. The final survey results show that the user preference rate of their product is close to 100%. They are full of confidence and think the product is perfect. But who would use your product every day if they don’t like it? (This example is similar to the "survivor bias" trap)

Data lies also occur in Internet finance products. For a typical example, the investment conversion rate of channel A is 60%, and the investment conversion rate of channel B is 40%. From the above data analysis, the user quality of channel A is obviously better than that of channel B.

However, if we combine different dimensions, the conclusion may be that the user quality of channel B is better than that of channel A. For example, considering the dimension of amount retention rate, the amount retention rate of users in channel A is 10%, and the amount retention rate of users in channel B is 50%. This shows that the "wool-grabbing" attribute of users in Channel A is obviously heavier. Users will make an investment and get the initial investment reward and leave. Compared with users in Channel B, they can stay and continue investing after the initial investment. Therefore, the overall quality of users in Channel B is higher. In this case, we get a conclusion that is completely opposite to the previous one.

Therefore, when product operators do data analysis, they must take various interference factors into account as much as possible. An incorrect data analysis conclusion is worse than not doing any analysis at all. After all, once you fall into the pit, you still have to spend a lot of effort to get out, and the time and opportunity costs are too high.

Conclusion

In the flood of information on the Internet, many people fantasize about grabbing some secrets to become famous overnight. Various marketing and learning accounts with low gold content (of course, except for the account where everyone is a product manager ) also like to share the following types of articles, such as "How to make a certain H5 that will go viral on WeChat Moments ", "Four methods to teach you how to create a hit", "Seven details for writing copy that will earn you more than 100,000 views", etc. Many Internet operators are keen on learning and studying the various "visible" principles and techniques mentioned above, but lack sufficient attention to many "invisible" system contents. Data operation analysis is the top priority among the “invisible” content.

Looking at data, finding causes, and thinking of solutions are basically the daily work of an operations staff. In addition to the step of "thinking of countermeasures", it is necessary to combine the operations staff's knowledge and understanding of the business, users, and products (the decision is usually made by the operations director ). The other two steps, knowing how to "look at data and find reasons", are essential skills for a qualified operator. The more proficient in this skill and the more you know how to use data to guide operations, the higher the rank of the operator will be.

Finally, I hope everyone can become an operations master.