Information flow promotion strategy to improve conversion

Information flow is now widely used in e-commerce, social networking, information and other fields. The form of information flow highlights the immersive experience, so the question is, how to improve the user conversion rate when they are in such a leisurely "shopping" state?

Growth is a hard indicator of today’s business, and designers cannot avoid having to bear such KPIs in this environment.

This common information flow, also called Feed, was originally redefined by Facebook in the social industry as News Feed. It is now widely used in e-commerce, social networking, information and other fields. The information flow highlights the information of the cards, and users can pull down to refresh it unlimitedly. It focuses on an "immersive" experience, and users can "browse" in it.

So the question is, when users are in such a leisurely "browsing" state, how can we improve the conversion rate and make users click in the feeds? There are several points that you can try and throw out for discussion.

1. An attempt to implement the “Thousands of People, Thousands of Faces” mechanism

What is “Thousands of Faces for Thousands of People”? Literally, it means that a thousand people see a thousand faces, and each person sees different content, thus achieving " personalized " customization.

For example, as a soft girl, your Taobao feeds look like this, with more beauty, clothing, and home products.

Imagine this scene: if the recommendations in your Feeds list are some mechanical keyboards and gaming equipment, then as a user, you will think, "Oh no, these things can't make me thinner or more beautiful, so what do they have to do with me? Scroll down and see if there is anything I can buy."

If you scroll 2-3 more screens and still see content that has nothing to do with you, such as "this year's popular POLO shirts", you may not have the patience to read on. Sorry, you will just exit.

The "Thousand Faces for Thousands of People" mechanism can solve this problem. Its goal is to provide each user with content that they like.

So how do we achieve having different faces for different people?

(1) The impact of the thousands of faces is based on the labels of user groups

The finer the tags are, the more the traffic will be segmented and the more accurate the recommendations will be. For students with label thinking, the display value and visitor value utilization rate are higher.

For example, if you are moving recently and looking at luggage packing tape on Taobao, you may be categorized by the platform under the label "moving". If it is more detailed, the "packing tape" may also be a label.

Then in your feeds, you may see recommendations for commonly used moving items such as packing tape, cartons, tape, packing rope, etc. Isn’t that very user-friendly? You may find something in these recommendations that you didn’t expect but can be useful.

In this way, personalized services can be developed based on consumers' browsing history and purchasing habits, and consumer groups can be labeled by analyzing this information, so as to achieve the goal of accurately recommending products to consumers. Concise information can meet the needs of consumer groups in a timely manner and help consumers quickly find content of interest, thus bringing an excellent user experience.

(2) The mechanism of "one thousand faces for one thousand people" is a recommendation mechanism

One is to make matching recommendations based on the behavior trajectory of C-end consumers (such as the browsing and clicking behavior of users on the page) and the shopping intention reflected by the channel, such as the example of "looking at the luggage packing tape" mentioned above;

The second type: Based on the B-end store (ie, the merchant), a store population portrait is set in the background (ie, some options are set in the background to tell the platform who its target users are), and the platform conducts intelligent matching recommendations. The information between the C-end and the B-end is interdependent, which constitutes the current complete Feeds "one thousand faces for one thousand people" recommendation mechanism.

At this time, what can interaction designers do?

(1) Structuring and componentizing the cards of the information flow

Card information is used to express the user's label information. Define the most overall framework and how to display various situations where information is missing. Please be patient and scroll down as shown in the picture below for a detailed explanation in the next section of this article.

(2) Stratify user groups and then match them with card information

There are multiple dimensions for stratifying user populations. The most commonly used method is to divide users into new customers and old customers. Occasionally, three more levels will be added: quasi-new customers, zombie users, and lost users. But the definition of new and old customers is different for different products, stores, and even locations.

For example, in a milk tea shop next to a residential area in Hangzhou, old customers can be defined as “people who have placed an order in this store within a month within a radius of 3 kilometers,” and new customers can be defined as “users who have never placed an order in this store within a radius of 3 kilometers.”

If this store provides delivery functions on major food delivery platforms, then for old customers who are labeled as "within 3 kilometers" and "have placed an order", discounts can be offered to old users, for example, 12% off on orders from old users, etc.; for new customers who are labeled as "within 3 kilometers" and "have not placed an order", a trial price can be offered to new customers, for example, "1 yuan trial price for new customers" and other methods can be used to attract them.

This stratification of user groups determines whether our information is accurate and effective. If new customers are defined as "users who have never placed an order in our store within a 1 km radius", we may lose coverage of customers who are a little further away because the distance is too close.

At work, such population definitions are usually decided after consultation with product managers and operations staff.

Second, the exposure of interest points

Interest points are factors that can influence C-end users to make decisions. These factors are beneficial to users. These interest points mainly include the following two aspects.

(1) Product information

Information on discounts (e.g. 20 yuan off for purchases over 200 yuan, 9.5% off for 88VIP), promotional information (e.g. 61 Carnival), order volume, user reviews, rankings, etc.

As shown in the figure below, there are various labels:

(2) Merchant information

Brand labels (such as brand, high-quality merchants, etc.). The disclosure of merchant information is a favorable factor for users who pursue brands. As shown in the figure below, the merchant’s “Brand” label.

Having such interest points alone is not enough. What designers need to do is to express this information. How to express it? The design scheme structures and components the cards. What is structuring and componentization? As shown in the previous figure (the product card structure of a certain platform)

The same card should have the same content and nature appearing in the same position, which means it is structured. As shown in the product media display area above, product information is displayed, which can be multimedia displays such as pictures, videos, live broadcasts, etc.

Componentization means that card information does not have to be complete. It can only support some of the information for display, and the missing information can be hidden. For example, if the product is not discounted for the time being, there will be no information in this area and it can be hidden.

As shown below:

3. Exploration of Interest Points

When users keep browsing but do not take action, it may be because they do not see the content they are interested in, so the strategy can be changed. The scope can be divided into the following three levels from small to large.

(1) Try to recommend different brands of similar products

Try the brand and see the results as shown below:

You can also try to further promote similar products through rankings or lists , both of which represent quality and recognition.

(2) Try to recommend products from different categories

In order to make the information that users see in the information flow richer, you can try to insert some other content information in addition to the recommended merchant tag content in the information flow. For example: You are a beauty lover, but you may also be a music lover. In that case, wouldn’t it be attractive to recommend some new headphones of the season in the cosmetics information flow?

Or maybe you are a straight man who likes sports and also likes playing games. Wouldn’t it be a happy thing to have these two products appear in your feed at the same time?

(3) Recommendation of similar content

During the browsing process, similar words can be recommended based on the user's browsing behavior, as shown in the figure below. There are two types of similar word recommendations: one is direct word recommendation, and the other is content recommendation based on user click behavior, both of which can serve to expand content.

As shown below:

In summary, to improve the conversion rate of Feeds, the logic is as follows:

• Do a good job in displaying and matching information. Structure and componentize the cards, and perform user stratification to match interest points.
• Real-time content recommendation changes based on user behavior.

Author: Sophiallg

Source: Sophia's Linglong Pavilion,

Related reading:

Understand the essence of information flow advertising strategy in one article!

Analysis plan for information flow advertising channels!