Half a year ago, due to coincidence, the author had the opportunity to design an e-commerce product from scratch. Because I am not an e-commerce person, I have no experience in designing e-commerce products, so I can only learn while doing it. Over the past six months, I have read a lot of books, videos and articles, and also communicated with many predecessors of my peers. Coupled with my own polishing in practice, I can now tell a thing or two in general. Taking advantage of some recent time, I am thinking about summing up the experience in the past six months and communicating with my peers.

as everyone who does e-commerce knows, there are three core modules of e-commerce products: information flow, capital flow and logistics. Because I am doing virtual goods trading, I don't involve logistics module, so I won't discuss logistics related content. The information flow can be subdivided into commodity information flow and order information flow (order flow is composed of information flow and capital flow). Therefore, I will explain it in three parts:

Let's talk about commodity information flow first in this article.

"commodity information flow" sounds very abstract. Regardless of the definition of the term, let's think about it first: what is the core of the product manager's concern in the whole process, from the time when many commodities are displayed on the website by the seller to the time when the buyer sees them and then makes a purchase? I think it should be the following two points:

To answer these questions, we should start with the design and construction of the whole commodity system.

when your product is very small, all the products should be displayed directly, and there is no need to classify them. For example, when Taobao was first launched in 23, there was no classification, and all goods were displayed directly.

when there are more and more commodities, it is inconvenient for users to find them, so it is necessary to have classification. In the field of e-commerce, we call this classification category, and the simplest one is the first-class category, such as Xiaomi Mall:/

As you can see from the above figure, there is only one category (first-class category) for each commodity, and there is no subcategory, and all the commodities under this category are linked under this category.

When the number of commodities goes up again, reaching thousands, tens of thousands, or even more, the first-class category can't meet the demand. At this time, the concept of multi-class category appears, which is what we call "category tree". Category tree is generally about three levels, and try not to exceed five levels. Because e-commerce has a recognized iron law called "funnel model", that is, the deeper the level, the greater the loss, just like a funnel, the smaller the mouth, so the category level should not be too deep.

The picture above is an example of a three-level category. When the seller uploads the goods, he needs to choose one level at a time until the leaf category is determined.

when the magnitude of commodities reaches millions, tens of millions or even billions, new problems appear again. For example, clothing can be divided into men's wear and women's wear, men's wear and women's wear are divided into T-shirts, pants, etc., while T-shirts are divided into many brands, and pants can be divided into nine-cropped pants, seven-cropped pants, etc. According to the length, such a category tree has been divided, and overlapping and overlapping are inevitable, which has become a network that is difficult to manage.

So when there are more and more commodities, the classification is more and more detailed, and the user search is more and more personalized, it is no longer possible to meet the needs of commodity management simply by category tree. At this time, another dimension classification method called "attribute" appeared.

How to understand "attribute"? Let's take a look at the following picture:

This is a description of the commodity "pants". We can use the adjectives on the left to describe it. This is the label we usually say. However, the classification of labels is too fine, and it is difficult to manage when there are too many labels. When we classify labels in the way on the right, these category names become what we call "attributes", and the labels on the left are what we call "attribute values".

for a popular example, we usually use wechat, and the contacts in the address book are sorted according to 26 English letters, which is like the category we mentioned above. Then we classify the contacts according to their different characteristics, such as "family", "high school classmate" and "college classmate", etc. If you like, it is ok to divide them into "ex-boyfriend" or "ex-girlfriend". These artificial classifications are labels, and classifying labels of the same nature into label groups is attributes (for example, classifying "high school classmates" and "college classmates" as "classmates"). Here is just an example. WeChat only supports tags, not tag groups.

it should be noted that when entering goods in the background, the attribute must be linked under the leaf category. For example, clothing-women's wear-miniskirt, which belongs to the category of leaves, can be linked with attributes, such as red, so that searching for red miniskirt can directly reach the goods. However, if you want to attach the attribute to clothing-women's clothing, there are many categories under women's clothing, so it is meaningless to attach women's clothing directly to red.

the "finding steel mesh" (/) in the above figure is a typical example of "category+attribute". Steel products are classified according to product name, material, specification, steel mill, etc., and then classified according to brand, etc.

does "category+attribute" solve all the problems of commodity classification?

of course, the answer is no. Let's take a look at a very common scene:

The essence here is that one product and one set of logic can't satisfy two completely different user groups well. So what should we do?

The first person who thought of the solution was a product manager of Taobao in 28. Once he visited Wal-Mart, he carefully observed the logic of commodity classification in traditional supermarkets:

Inspired by this, he came up with an architectural design scheme of "foreground category+background category"-dividing a product into two parts, one to satisfy the buyer and the other to satisfy the seller, that is,

the original one. The leaf category of any foreground category can correspond to any one or more background categories, and it is not necessarily a background leaf category. For example:

Let's take a look at an example of an actual scene. Taobao (/):

The red circle in the above figure is a typical foreground category displayed by operators for operational needs, which maps specific goods under certain categories or attributes in the background.

This design has established the product classification system model of most of our current e-commerce products: foreground category+background category+front-background mapping management+attributes

Looking back, from the launch of Taobao, China's first e-commerce website, in 23 to the blossoming of e-commerce websites, the evolution path of the product classification system can be summarized into five steps. As shown below:

At this point, the first knowledge point I want to share with you-commodity classification system has been finished. In view of my limited experience, there may still be many places that I don't understand, or some places that I can't express clearly. Welcome to clap bricks ~

I hope that your sincere communication with me will spark wisdom ~ _ ~