How does Pinduoduo use the price-cutting “strategy” to gain 100 million traffic?

In the first half of 2018, we made a prediction:

Due to the rising labor costs of low-end Internet talent and the overall shift of the industry's competition focus to data- and strategy-driven refined operations , we believe that in the next 2 to 3 years, "strategy" talent will become a common and rigid demand in the entire Internet industry.

The phenomenon of relying on a few "strategies" within a product to eliminate a lot of human labor costs and greatly improve efficiency will become more and more common.

Therefore, we believe that "strategy" ability will be the most worthy ability for Internet practitioners to learn.

However, so far, we have also found that the word "strategy" is still abstract and vague to most Internet practitioners. Many people do not know what it means. Even if some people understand the meaning behind the word "strategy", they do not know how to learn it.

Therefore, we hope to popularize the following through a series of real case studies in a regularly serialized form: what exactly is strategy, what problems it can solve and what value it can bring, and what is the underlying thinking method behind it. We hope that more people can enter the door of "strategy" through this.

In the first article of this series, we will focus on Pinduoduo, which has been at the center of controversy recently. We will explain to you the application of "strategy" in it by combining one of its core functions - bargaining for free.

one

Let’s first understand what “bargain and get it for free” means.

In short, this is a function in which Pinduoduo uses the bait of "free access to various goods" to induce users to help bargain for them by sharing links to groups or WeChat friends, thereby continuously obtaining traffic within the WeChat ecosystem . Although we are unable to know the relevant data behind this function yet, judging from the widely circulated related news and discussions on WeChat, Zhihu, and Baidu, it has undoubtedly become another traffic killer under Pinduoduo.

Let’s take a quick look at a few screenshots to understand the main usage scenarios and processes of the “Bargain for Free” feature, as well as its presentation form after being shared on WeChat and QQ:

After entering the Pinduoduo homepage, you can find the entrance to "Bargain and Get It for Free":

After selecting a product you want, you can bargain for it yourself first, and then Pinduoduo will prompt you to share the link with your friends to help you bargain.

The style of sharing the bargaining link to the WeChat group is shown in the picture above.

It is worth noting that Pinduoduo’s “bargaining” to get something for free has several core mechanisms:

Once you start bargaining, if you don't find someone to bargain down to 0 within 24 hours, you will lose the opportunity to get the product for free.

The same friend can only help you bargain for a product once.

The same user can only help three friends bargain every day to prevent a small circle of people from helping to bargain back and forth.

This means that even if you have found dozens of friends to bargain for the product you want down to just 1 cent, you can only continue to find people to help you bargain, otherwise all your efforts will be wasted.

It is driven by these mechanisms that countless " Pinduoduo bargaining mutual aid groups" have emerged. Pinduoduo's "bargaining" mini-program links have become popular in WeChat groups and QQ groups, showing a strong "fission" effect, and has become another traffic killer in recent times.

Of course, just like the Pinduoduo product itself, there are also many controversies and chaos behind the “bargain and get it for free” strategy. Here, we might as well put aside all the controversies and chaos, focus on the product and operation perspective, and specifically explore the more interesting aspects behind this "traffic killer" that are not easily discovered.

two

From the perspective of product function design, if we want to sort out the main product process of "bargain and get it for free", it is generally like this:

This process seems to be quite clear. However, if you are the product manager or operator in charge, even with such a clear process, can you ensure that this function will become a "traffic killer" and will not encounter other problems?

not necessarily.

Here is a common sense: the more uncertain variable elements there are in a business process, the lower the efficiency will inevitably be as the number of users it carries increases.

For example, if 100 people of similar age, income, family status, etc. participate in this event, we may rely on the clear process and rules mentioned above, plus a few products that meet the needs of these 100 people, which should be enough to basically ensure the success of this event.

But what if we need to carry 1 million or even 10 million people?

At this point, you will find that at least a few issues will greatly affect the final effect of this activity:

First, as the user base grows, user needs will inevitably become more diverse. So, how to solve the matching efficiency problem between "products" and "users" at the scale of 1 million users? In other words, if there are 50 items participating in the "free" promotion on your website, different users will definitely be interested in different items. So how can you allow as many users as possible to quickly see the items they are interested in, so as to stimulate their willingness to participate in the bargaining?

Second, in the “bargaining” process, for different products, how many people do I need to help me bargain in order to successfully “bargain to 0”? And should the amount that users in different status bargain for be different, so as to maximize the probability of users successfully participating in the bargaining and being willing to invite others to help them bargain?

Third, how can we enable those users who help others bargain to become the initiators of “bargaining for free” as much as possible, so as to maximize the “fission propagation” effect of this mechanism?

These three problems would be almost impossible to solve under the traditional state where business processes must be driven by manpower.

At this time, you may need to rely on "strategy".

three

In short, the so-called "strategy" is often a set of rules and mechanisms driven by data and existing in the product.

To explain it from another perspective, if the core business of a company is highly dependent on human work, employee management issues may be extremely complex and stability may be difficult to ensure; however, if the core business of a company relies on machines to run, then it will be very stable and sustainable. The operation of the machine needs to be constrained by rules, and the rules that support the operation of the machine are strategies.

When we are faced with massive demands and information that need to be processed, and have accumulated enough data, we can definitely use strategies to help us solve many problems that are difficult to solve by relying on manpower.

As for effectively formulating a "matching" strategy, the most basic thinking logic is often like this:

First, use data mining and analysis to classify users or the solutions you provide to users through some logic (for example, Pinduoduo users can be divided into price-sensitive and quality-sensitive types, and goods can be divided into different categories. Different categories can be further divided into low-priced, medium-low-quality goods, medium-priced, medium-high-quality goods, etc.);

If an existing solution cannot satisfy all users, more different solutions should be designed, provided that each solution can satisfy a typical user.

Matching between different solutions and different users is accomplished through a set of judgment rules.

For example, if a user searches and browses "electric kettle" multiple times, we can automatically tag him/her with a label related to electric kettle through the machine. In this way, when he/she enters the "bargain and get it for free" mode, we can prioritize him/her with opportunities to get electric kettle for free through bargaining. This will significantly improve the efficiency of his/her participation, and the entire process does not require any human intervention and is completed entirely by machine.

Basically, Pinduoduo also formulates corresponding product and operation strategies according to this logic for the three problems mentioned above, thereby greatly improving its efficiency. Let’s look at them one by one.

Four

1. Product selection strategy

First of all, how do you decide what kind of products should be pushed to users?

Traditionally, there are generally two types of product recommendations on e-commerce platforms.

One is the recommendation in a fixed position, such as the homepage banner or some fixed recommendation positions. To determine the products recommended in a certain position, most of the time, they are selected by operations staff, and the schedule and resources are determined in advance.

The other is the recommended position that changes according to certain fixed rules. The most common one is the so-called sales ranking list . There may also be some content-based recommendation methods, such as Taobao headlines.

Basically, most of the time, the products everyone sees are the same and there is no particular difference.

In this process, not only is the conversion rate uncontrollable, but a large amount of manpower is also required to complete this task.

Policy-based recommendation strategies are completely different. Through strategies, we are able to form customized and product recommendation solutions based on complete personalization.

Combining the way of thinking we just mentioned, let’s imagine a few scenarios and try to restore how the product selection and recommendation strategy is formulated. It should be noted that the scenario we are going to describe next is only based on some logical assumptions, and the actual business logic must be much more complicated than this.

First of all, what we need to do is to classify users or the solutions you provide to users through data mining and analysis using some logic.

We assume that users who use Pinduoduo's bargaining function can be divided into two categories based on whether the maximum single transaction amount reaches 500 yuan. Those who reach more than 500 yuan are marked as price-insensitive, and those who do not are price-sensitive.

At the same time, we use whether the unit price of the goods exceeds 50 yuan as the standard. If it exceeds 50 yuan, we define it as a high-priced product, and if it does not exceed 50 yuan, we define it as a low-priced product.

After users and products are classified and labeled, the next step is to match users and products. In theory, there will basically be a matching relationship between users and products.

However, if there is only this kind of strategy, it is obviously not enough. Some users may only be sensitive to the prices of some products, but not to the prices of other products. If they are always pushed products with higher prices, or vice versa, it will cause many problems.

At this time, we need to continue to design more solutions through data mining and analysis.

For example, we use whether the same product has been viewed more than three times as a criterion to divide users into target users and non-target users of the product. At this time, users are divided into four categories:

At the same time, we also need to classify product types. For example, the products can be classified according to their types, such as electronic products, daily necessities, clothing and accessories, etc. At this time, goods are divided into more categories.

At this time, some new rules emerged:

For target users of a certain electronic product, if the highest single transaction amount does not reach 500 yuan, we will push more electronic products with a unit price below 50 yuan to them, such as data cables, power plugs, etc.

For target users who also purchase electronic products but whose maximum single transaction amount reaches more than 1,000 yuan, we will push more high-priced products to them, such as mobile phones, computers, headphones, etc.

In the process of continuous exploration and experimentation, matching rules for different users and different products will gradually be formed and used in different scenarios.

Following this logic, after the rules are perfected, a possible scenario is that a user searches and browses "electric kettle" many times and has never made any purchases on Pinduoduo. The machine will automatically label him/her with a label related to electric kettle and a price-sensitive label. When he/she enters the "bargain and get it for free" mode, we can give him/her priority to push opportunities to get electric kettle for free through bargaining. This will significantly improve the efficiency of his/her participation, and the entire process requires no human intervention and is completed entirely by machines.

Of course, one point that needs to be reiterated is that the scenario we described above is only a logical deduction. In actual work, the process must be far more complicated than the scenario we described.

For example, at the user level alone, we can directly consider labels such as whether anyone in his first-level social network has purchased a certain type of product, how many times he has purchased a certain type of product, under what circumstances he would give a bad review to a certain product, etc. At the product level, there are also many tags that can be added.

In each dimension, only one label needs to be added, and the workload of matching them will increase exponentially. This is almost impossible to accomplish with human power, but through repeated data verification, continuous testing, and ultimately coming up with a rule that works best, a machine can accomplish it easily.

This way of thinking also applies to the question of how to determine the amount of bargaining.

2. Fixed knife strategy

Obviously, the number of bargaining attempts will have a great impact on the spread of the bargaining function.

If you cut it too quickly, it will be gone in a few cuts, which will not only fail to promote sharing, but also make the activity cost very high.

If you chop too slowly and fail to succeed even after dozens of cuts, many people may not be willing to do it, and this function will be worthless.

This is also a problem that can be solved by strategy, and the way of thinking is actually no different from just now.

First of all, we need to classify the groups of people and solutions.

For example, we can divide users who participate in bargaining into two categories, new users and old users.

The solution here is no longer the product, but the number of bargaining for each product and the amount of each bargaining. These categories also need to be determined through data separately.

Here we assume that a new user can bargain 20 times for the first time, while an old user needs 30 times to complete the task, and the bargaining amount only needs to reach 100% in the end, and the process can be random.

Now we have two clear rules, which are enforceable, but clearly not sufficient. There are several reasons for this:

Users will quickly discover this pattern and find a fixed bargaining group of 30 people. The large amount of bargaining will result in high bargaining costs.

Secondly, if there are only these two rules, then the function will definitely not be contagious, because the 30 people in the group have no motivation to find other people.

Moreover, this actually discriminates against old users to a certain extent, and users may not be willing to continue using it.

Therefore, in practice, it is necessary to set different bargaining numbers and bargaining amounts for different users.

For example, for users who initiate bargaining, they may need to cut the price to a very high level in the first attempt, so that they can share the power and feel that this feature is worth the money.

For example, for users who are participating in bargaining for the first time, they may need to offer the highest possible amount so that they will be more receptive to this feature. Moreover, users who initiate bargaining will be more likely to share it with more new users.

In actual experience, Pinduoduo has indeed formulated rules based on this, for example:

New users can bargain for about 30% of the total amount;

Users who participate in bargaining for the first time can bargain down to about 20% of the total amount;

Old users who have participated multiple times can bargain for very little, sometimes even for 0 yuan.

Under this rule, users will be more motivated to constantly look for new users who have not participated in bargaining, allowing product features to spread and grow at a rapid speed.

These are just a few rules for matching user tags with bargaining amount tags. Other rules, such as matching user tags with bargaining amount, matching product tags with bargaining amount, etc., actually need to be determined through strategies, but we cannot perceive them in a short time. These strategies are important driving forces for the rapid growth of bargaining functions.

The same way of thinking can actually be applied when facing the last question, that is, how to maximize the "fission propagation" effect of this mechanism.

3. Recall Strategy

If you have experienced Pinduoduo's bargaining function, you should have noticed that after you help your friend bargain once, the system will automatically start to push various bargained products to you for you to "get for free."

From a functional perspective, it is an ordinary push recall function.

However, Pinduoduo has also tried its best to make the products it pushes different for different people.

In this process, there are many factors to consider, such as whether you are the target user of a certain type of product, whether you are price sensitive, whether the sales volume of the product is high enough, and whether the price of the product is low enough. These are all factors that need to be considered.

In addition, factors such as what products you helped your friend to buy and your friend’s preferences (because your preferences are likely to be similar) are also taken into consideration.

This process, by clarifying some rules and strategies for matching, forms a whole personalized solution, and then through personalized push methods, it can make users more likely to return to the bargaining function and achieve fission growth.

The entire thinking process and thinking path are actually the ones we just mentioned.

I wonder if after reading the above analyses, you will have a clearer understanding and cognition of how strategies are formulated and how strategies play a role in the bargaining function.

Basically, the way of thinking we just mentioned is universal when faced with product features that need to be "matched" and drive growth through matching. It is not just Pinduoduo's bargaining function, group buying function, and the push mechanism of products such as Toutiao and Douyin that we are already familiar with. The underlying foundation is the above methodology, and by applying similar ways of thinking, we can all formulate effective rules.

From a broader perspective, in the Internet industry, in theory, when any product reaches a certain user base, there will inevitably be some problems that need to be solved through strategic means.

In other words, in the current Internet industry, no matter what product you are making, strategy is everywhere.

For example, the following typical business problems and scenarios are where “strategy” in the Internet industry is most often used:

As the number of users of a product increases, it is necessary to consider refined operations;

A product may involve functional modules such as search, recommendation, and supply-demand matching;

A product has large-scale transaction and closing data, and needs to improve the closing rate and optimize the profit margin;

A product needs to consider driving user growth through data;

A product in its growth stage needs to consider using subsidies to drive user growth on a large scale, and hopes that its subsidy methods and forms can become as efficient as possible;

…

As a product manager, if you can master the ability to formulate "strategies", you will most likely be able to break out of the vicious circle of relying on "manual labor" and start to rely on "mechanisms" and "machines" to help yourself improve business efficiency; at the same time, you will be able to solve more complex problems than before, making yourself more valuable and finding better career opportunities for yourself.

Similarly, as an operator yourself, if you have the ability to think about "strategies", you are more likely to standardize your operational actions and achieve work that relies on "machines" for refined operations, helping you to face challenges on a larger scale and in more dimensions and become a more irreplaceable presence in the market.

Therefore, we say that "strategy" ability will be the most worthy ability for Internet practitioners to learn.