Traffic distribution logic of Internet finance products!

When making products, everyone likes to talk about data-driven. The essence of data-driven is to use data to reflect traffic conversion efficiency and guide the improvement of traffic conversion efficiency.

The proposition of improving efficiency is to find the optimal solution under specific limited conditions. In essence, all traffic problems are advertising problems. In the end, it all comes back to bidding advertising. The core is that the top advertising space has the most expensive traffic (value/price), and the long-tail traffic can be more efficiently utilized through precise allocation. The example used to explain traffic distribution today is an Internet finance product that buys and sells traffic. It is the most representative traffic monetization product besides advertising platforms.

The core of this product is to find traffic, allocate traffic, and convert traffic.

01 Find the most suitable traffic

Only by finding the most suitable traffic can we increase the traffic level from the beginning, make full use of the traffic budget, and save time on trial and error.

In order to find the most suitable traffic source, you should go through the following steps:

[Set target population - target population attributes - find channel types - compare channel values ​​- channel drainage - channel value assessment]

Take Huanbei, which has been advertising crazily on Zhihu recently, as an example:

• Target group: (with credit cards, willingness to spend greater than ability to spend) young people who have just graduated and are working, workers in big cities
• Target group attributes: those who have access to abundant information and have financial ability
• Look for channel types: content platforms, e-commerce platforms, Internet finance platforms...
• Compare channel values: Query prices across various channels and find the channel with the highest estimated ROI
• Channel diversion: Develop advertising plans based on channel characteristics. If you use direct H5 to collect user mobile phone numbers, you can switch and adjust them in real time in the background to improve conversion efficiency.
• Channel value assessment: Based on the user behavior performance of each channel, establish a channel value assessment model and establish channel clearance standards. I will talk about this later when I have the chance, so I won’t go into details here.

02 Allocate traffic to the most effective entrance

Traffic has come to the product, in order to make full use of the traffic and create the highest value. Most products are not limited to one method of monetization. For example, Internet finance products may have functions such as bookkeeping, credit card management, investment, loans, community, and credit checking. Some of these functions are designed to increase user activity, some are designed to obtain user information, and some are designed to monetize.

Among these functions, allocation can be made according to strategy/ROI. Layered and graded traffic diversion improves traffic efficiency.

For the main product, the responsibilities are as follows:

• Ensure the availability of core functions
• Collect user information and classify users
• Direct traffic to each branch function
• Providing operational tools

The key here is to improve the efficiency of initial traffic distribution. For example, users who invest small amounts and users who invest large amounts may not overlap (just an analogy, not necessarily in reality), so it would not be efficient to not split the traffic and let it flow to these two functions on its own. User types should be analyzed at the beginning, and users should be directed to corresponding functions.

03 Improve final traffic conversion

The third step is the final traffic conversion.

There are several steps to convert traffic into cash. For example, in mutual financial products, the process is as follows:

The user enters the homepage - the user enters each functional area - the user clicks on a loan product - the user enters a mobile phone number to register - the user applies for a loan - the user checks the loan progress - the user completes the loan - the user is guided to other products for secondary monetization.

This process is a large funnel model, and there is room for conversion improvement in each step. I will write about the conversion improvements between each step bit by bit when I have the chance.

At the same time, the traffic price of Internet finance products is very expensive, and the traffic conversion efficiency must be improved to cover the cost. When guiding customers to various loan products, if the loan products are self-operated, then it is necessary to consider the ROI of each self-operated product (which can actually be regarded as opportunity cost). If the loan products are not self-operated, it is necessary to consider the bids of these Party A products. From this perspective, this model is close to bidding advertising, where the best position and the most traffic will be directed to the highest-priced products.

In the short term, this strategy is the best, but it will result in only 2-3 super clients on the platform (even conservatively estimated that the top products will absorb more than 50% of the traffic). If you think about it more extremely, if the client's costs allow, it is optimal in the short term to direct all traffic to one client. However, this will cause the product to be overly dependent on a few parties, which is extremely risky. At the same time, it is impossible to contact new parties and find higher-priced products.

This leads to new questions:

• Traffic distribution between operation positions and product lists
• If we make scientific traffic distribution between products
• Coordination of ROI with other factors: such as customers, etc.

These developments are a larger article, with two core points:

• First, control the interaction, such as appropriately controlling the display of the operation position and the screen ratio;
• The second is to control the product mechanism.

So now assuming that we have solved the above problems, new problems arise.

Different locations attract users of different qualities.

From my personal experience, the traffic generated by operation positions like banners is very poor, and many people just click on them casually. This can also be extrapolated to many traditional operation positions. However, the head position of the product recommendation list is very good, but the tail position of the recommendation list is very poor.

Although we have solved the problem of "quantity" distribution above, the problem of "quality" now arises again.

How to distribute homogeneous traffic reasonably?

This involves flow control.

• Control time;
• Control weight level;
• Control device number/mobile phone number/IP···;
• Control scene (display conditions);
• Control the customer base.

By controlling the quantity of products in various ways, we can achieve maximum quality control in the short term. Reaching this level can basically meet the needs of quality control and quantity control. But could it be better?

This requires analysis of users and user behavior. To analyze users, we first compare them with our self-built user database, and second, we obtain user behavior to conduct risk control analysis. Users of different qualities are sold to different customers at different prices to obtain maximum profits.

at last

The core of traffic control in the early stage of the product is to focus on the big and let go of the small, while in the later stage it is refined operation. In the early stage, we need to establish a system, start from the big picture, and improve the overall data. In the later stage, we need to focus on small details, identify abnormal data, and formulate corresponding strategies.

Author: Zhang Xiaosi

Source: Product Monster