Data-driven operation process for Amazon sellers

November 15, 2017 20:10 (China Time)

Host: Ye Shen

Topic: The entire process of data-based operations

Location: WeChat group of "Knowing Everything" Wendao Study Group

Mode: Voice/text live broadcast

「Activity content record」

Hello everyone, I am Ye Shen from Longhua. Today I will mainly share some thoughts on data-driven operations. Please forgive me if I say anything wrong.

In my opinion, whether a company can grow and develop depends on how strong its core competitiveness is. Different companies have different components of their core competitiveness. Some rely on business models, while others rely on strong capital. So, if the business models are the same, in which aspect does the comparison of the strengths of the companies manifest?

Specifically in the cross-border e-commerce industry, we can see that there are roughly two company models: one is a trading company, most of the companies are trading companies; the other is a product company, which is a factory that focuses on producing a single category of products or has its own core research and development capabilities.

How can these two types of companies build their core competitiveness?

We won’t talk about product companies. There are many examples to refer to when making products, and the threshold is also the highest. We focus on trading companies, which account for the majority of this industry. For trading companies to do Amazon, it is nothing more than buying low and selling high. Find a way to purchase a low-cost product, and then sell it through the Amazon platform. If a trading company only opens a store on the platform, it will not be able to accumulate its own customer base, and it will be difficult for it to have a say in the channel. In other words, if you only sell goods on the Amazon channel, even if you sell a lot, you still cannot build your own channel. Because you don’t know when the platform policy will change, and you can’t control it.

On this basis, since trading companies cannot control the product side like product-oriented companies, nor can they control the front-end channels, what can they use to build a moat for such companies?

My answer is standardized and process-based operations or operating models.

Standardization means that each process is executed very carefully. The execution of these processes is like having a set of detailed instructions, and you just need to follow them. Standardization means replicability and mass production.

Why is this standardized model suitable for trading companies to join Amazon or enter the cross-border e-commerce industry?

Because trading companies are destined to have a short product life cycle, they are destined to promote one product after another, over and over again. This repetitive process is the process of replicability or mass production. Since mass production is required, it means that it must be standardized, otherwise the unit cost will be too high and it will eventually be eliminated by the market. So, how can trading cross-border e-commerce companies achieve standardized workflows?

My idea is to build data models or detailed working modes one after another, and standardize each workflow through these models. Based on my understanding of trading companies that mainly use the Amazon platform, I divide the processes of such companies into four major parts: product development or product selection, data-based daily operations, data-based procurement and inventory management, and data-based promotion.

We need to establish our own standards for each of these four areas. This standard is like a factory assembly line that can produce products in large quantities. In other words, this standard can withstand the promotion of one product after another. When it comes to each process, I have some thoughts as follows for your reference (I don’t have much experience in purchasing and inventory management, so I won’t expand on this aspect for now.)

Part 1: Product Selection

How to digitize product selection? In addition to investigating the offline market, developers or product selectors usually use some online data to determine whether to enter a certain market segment.

Since each trading company has different resources and is familiar with different categories, it is necessary to establish its own product selection criteria based on its own company resources. Based on this standard, with the help of relevant tools and techniques, it can select suitable products for sale. In this process, each company must have its own set of data models and its own operating procedures. Let me give you an example.

For example, we have decided to develop a product that is in high demand, with a medium level of competition, no monopoly by big brands, and a gross profit margin of more than 30%, and it must be in certain categories. Because the supply chain of companies in these categories may be relatively rich. After determining this standard, we will further refine this standard. Through this standard, we can establish an instruction manual similar to an operating manual, so that both new and old employees can carry out their work step by step. For example, it is stipulated that the monthly Google search volume of a certain keyword must be greater than a certain number, the search results of the keywords on the site must be less than a certain number, and the sales volume of the top 20 on the site must be no less than a certain number, etc. In addition, for judgments such as gross profit, ratings, and after-sales, we must establish our own accurate data models. Only in this way can we establish a standardized operating process.

Here is a specific example of an attempt I made to do this.

First of all, the categories are relatively certain, after all, we can’t focus on too many categories. In daily data research, we can see products with good trends from the new product list, soaring list, or vertical websites; or see related products, we can start to decide whether to enter based on our own standards. Generally speaking, we use core keywords to conduct research and estimate sales data.

Some time ago, a friend commissioned me to research the situation of dog nail clippers. He has pet factory resources and hopes to investigate whether the related products can be sold profitably on Amazon. At this time, we clearly know that the core keyword is dog nail clipper. After knowing the core keyword, through a few simple steps, we can know the market capacity, competitiveness, etc. of the core keyword. Here you can use some tools to achieve this demand, or you can decide how to obtain this data based on the situation of your own company. My method is relatively rough.

For the word dog nail clipper, I would use Chuangxiang Collector or Locomotive Collector to collect the data I want. After downloading, I would remove some duplicate or useless data and organize them into data that I can use as a reference. The following is the original data collected:

The following is the data obtained based on the above table and the Amazon platform itself. This table is borrowed from the survey form of a certain Internet celebrity (thanks for his selfless dedication).

Through these two steps, we already know the sales volume of this keyword in detail. Before these two steps, we will also use some simple methods to determine whether this keyword is worth further research.

Let’s take this word as an example:

1. Google Trends search volume comparison:

Here you can compare it with the keywords of the categories you often work in to get an idea of ​​the approximate search volume.

2. Number of search box results and comments/rankings:

The number of search results is currently not displayed on the PC side, you need to view it from the mobile phone or from the background.

3. Global store assistant sales tracking and keepa to check whether the ranking changes of a single ASIN are stable:

These indicators are mainly negative. For example, if the search volume is too large or there are too many comments, we may not conduct in-depth research on such keywords. After all, the competition is too fierce and the cost of promotion may not be affordable for the company.

Part II: Daily Operations

I divide daily operations into four parts: listing, off-site promotion, traffic and inventory monitoring, and PPC operations.

For these four parts, each company must have its own set of strict processes, which are precise and data-based, and we must reflect them through data. For example, when to put a product on the shelf, what to do, there must be a clear time point; when to carry out off-site promotion, how to cooperate, and subsequent effect analysis; how to establish traffic data for each sku, and combine PPC data for multi-dimensional analysis. How to design a reasonable inventory monitoring formula or system based on the company's capital turnover, etc. PPC operations, which reports should be analyzed and how to set the company's data requirements for conversion rate or return on investment, etc.

All of these have clear standards or data indicators, and operational adjustments must also be based on data indicators.

Daily operation data table:

Part 3: Data-based Promotion

This part mainly refers to how to promote products with the help of on-site and off-site PPC and make some adjustments through data after they are launched and optimized. We need to establish a data indicator system for these promotions so that the promotion can be within a reasonable and controllable range. The promotion also has clear steps and rhythms, which are clear and concise. Subsequent adjustments also have clear data basis.

These are my thoughts on the entire process of data-based operations. I hope they will be helpful to you. You are welcome to ask questions and discuss the issues together. The reason for this idea is to establish a set of standardized processes that can be replicated in large quantities, thereby building a moat for a cross-border trading company.

Thank you everyone!