The underlying logic of Pinduoduo store operations

This article starts with the problems caused by operating Pinduoduo. By analyzing Pinduoduo's product recommendation mechanism, the author organizes multiple assessment dimensions of Pinduoduo recommendations based on personal operating experience and explores the underlying logic of store operations. I hope it will be helpful to you.

I have been working on Pinduoduo for a long time, opened many stores on it, and promoted many products. I have been thinking about a question: why do my products with high bids not sell as well as those with lower bids? Why are my products priced so much lower than others, but still cannot sell as well as their flagship stores? Why do my links stall while I am promoting them? And so on.

I think about it and look for answers every day.

If you ask ten people, you may get ten answers . Which one is right? It is difficult to judge. The rules of Pinduoduo stores are like a black box, and you have to guess.

As the saying goes, if you keep thinking about it, there will be a response.

When I was operating a new link and looking at the data trend of this link, I suddenly understood Duoduo's recommendation mechanism. This is thanks to my previous platform operation experience and understanding of the Thousand Faces Algorithm.

I once worked as a distributor in an app market, with a platform with millions of daily active users. My only responsibility was to maximize the value of traffic. I had many cards in my hand, including popular mobile games, paid apps, must-have apps, etc. Some of them could be monetized, with varying degrees of monetization, while others could not be monetized but were in high demand.

I have limited resources. How can I schedule products to make more money? Do it manually? No, that would be too much work and it would be difficult for users to find the products they want. The best way is to schedule through algorithms.

At that time, the company was not big and there was no algorithm engineer, so I drafted a very basic algorithm formula myself, setting weight coefficients based on several dimensions such as the popularity of the APP, software size, monetization ability, conversion rate, etc. , and finally ranking the APP according to the final scores of each weight. It was simple and effective.

Later we witnessed the rise of Toutiao and Douyin. What was the core of their rise?

It's just that everyone has their own unique face. The quality of content is not based on the subjective judgment of the editor, but on the data of the content, such as the completion rate, forwarding rate, and like rate. Let good content automatically get recommended, and leave no room for junk content to survive.

Later, platforms including Taobao, Dianping, Meituan and others adopted this logic, as did Pinduoduo .

So if you look at Pinduoduo's product recommendation logic from a distance, it will be clear at a glance. Based on my personal operating experience, I have organized the dimensions that need to be evaluated for Pinduoduo's recommendations as follows: data score (about sales data), quality score (about front-end data), product score (about back-end data), and weight score (about store attributes) .

01 Data Division

The data points are the most basic sales and conversion rates , as well as collection rate, comment rate, repurchase rate, bounce rate, etc. The reason why I focus on sales and conversion rate is because I personally think that the weight of these two data is the highest.

If the weight of the collection rate is 10, the weight of the conversion rate is at least 10,000. Some links can be detonated in 7 days. You must know that there is no data in 7 days. The only data you can refer to is the conversion rate.

I guess the platform understands that as long as your conversion rate is high enough, it means everyone recognizes your product, it must be a good product, and I will definitely try my best to give you traffic. It is not ruled out that there are short-term operators who change the link every two weeks. Why? DSR has come out, and it can't be done anymore, but it doesn't prevent others from making money in one wave. It's all about the time difference of the algorithm.

The data points are assessed based on 30-day data, which gives us room to operate seasonal products. Taking summer products as an example, the detonation point is June, so we have to start driving in May, even if we lose money.

You should increase your sales first. When June comes, your sales weight will be at the top under the same category links. Your links will naturally have the possibility of becoming popular. If others want to catch up with you, the cost of investment will not be low.

02 Quality score

The quality score is indicated in the background. Every day when I open the background, I see that the linked product title needs to be optimized and the product experience score is less than 5 points. This is what everyone can see. For example, the main picture and detail score that cannot be seen can only be judged based on personal operation experience.

When I first started e-commerce operations, I had no idea about the details of the main picture. After doing it for a long time, I realized that the most basic point in making materials is to try to keep the style and color scheme consistent, there should be no more than 3 colors in one material, and don’t make complex pictures.

Why does a picture like this fail to pass the platform review? It’s not because it’s ugly, but because the effect of such a picture is not good. A product picture with a white background is better than a picture full of text .

There are standard tutorials for quality score. You can master it by reading more. It is not difficult. It is nothing more than the structure of the title, the elements of the main picture, and the content distribution of the details. Just follow the existing experience tutorials. The quality score is not a big problem. Just pay a little attention.

03 Product classification

You should pay special attention to the backend data. The backend data is not as intuitive as the frontend data. Sometimes your data drops, and you may not think that it is caused by the reduction of your DSR. After all, there is no obvious correlation between the two.

Here I can clearly tell you that if the product quality score is low, your link will have a great impact, especially for those who play the natural flow, DSR is your pass. From the perspective of the platform, the platform is not for you to fleece the sheep, it also wants user experience , otherwise why would it do what Qianfu (merchants) point out, only refund?

I have included DSR, refund dispute rate, customer service response rate, navigator score, etc. into the product score. The platform will definitely assess the user experience of your link (product) from multiple dimensions.

For example, if you sell cherries at half the market price, is your conversion rate too high? In the early stage, the platform will continuously push traffic to you. Then I bought a box and found that half of them were rotten, so I gave a one-star bad review. After most buyers gave a one-star bad review, the DSR fell like a waterfall, and the platform gradually stopped pushing this link.

Of course, according to the multi-dimensional assessment mechanism, if your sales are good enough, the platform will continue to promote you , but the cost of using extraordinary means to increase sales is not as cost-effective as creating a new link.

The speculation is as follows: for DSR4, your single-link daily sales must reach 10,000; for DSR3, your daily sales must reach 500,000; for DSR2, your daily sales must reach 8 million; and for DSR1, your daily sales must reach 100 million. The above data is only to explain the idea of ​​comprehensive scoring and is not for reference.

04 Weighted Score

My personal understanding is that store attributes correspond to different weights, flagship store, brand store, franchise store, and personal store.

When you search for a keyword, most of the stores ranked at the top are black label stores. It’s not because the data of black label stores is good, but mainly because the weight of black label stores is high enough. The weight of a link of your franchise store with sales of 100,000 a day may be the same as the weight of a black label store with sales of 10,000 a day.

Some people say that even black label stores can operate well, which is true . Because the link itself has a high weight, the requirements for everyone's operation ability are not high.

So my suggestion is that if you can apply for a flagship store, try not to open a personal store. After all, the platform will give extra care to flagship stores.

05 Last

This is the underlying logic of Pinduoduo. If you read it several times and understand it, you will know how to play Duoduo. In addition, there is another concept that you need to understand thoroughly, which is the traffic pool.

The comprehensive score of the link is a gear, which needs to be combined with another gear, the traffic pool, to perfectly generate a steady stream of income. When everyone has mastered these two parts, congratulations, you have graduated from Duoduo E-commerce.

Author: Tiger Talks Operations

Source: WeChat official account: "Tiger Talks about Operations (ID: laohujiangyy)"