The traps and blind spots of big data selection
3 min readMay 22, 2021

The current phenomenon of cross-border selection of products, basically every cross-border e-commerce company will have more than one selection tool. The phenomenon is that more and more companies use tools to select products, and people are increasingly relying on this tool selection. According to various filter conditions set, within a certain range, help sellers choose the competitor products they want to refer to.

At present, the reason for the prevalence of data-based product selection in the Amazon e-commerce industry is indeed because there are too many sellers. Under this high-efficiency and high-competition cross-border situation, everyone is in urgent need of copying artifacts, and products are urgently needed. More and more narrow, the phenomenon of product homogeneity is getting more and more serious, often a hot new model that has just been launched will be accompanied by a lot of the same models, and eventually it will evolve into a price war.

Then why copy money? Frankly speaking, it is indeed the safest choice to take the path that others have taken, sell products that others have already sold, and use lower prices, better copywriting, and more dazzling pictures to grab the market. It’s just that in the long run, this road should be narrower and narrower, because the industry shuffle is already obvious. When there are fewer and fewer sellers you can refer to, the more you rely on your independent innovation decision-making.

Market capacity under big data may be a trap

Under the concept of digital product selection, the most important indicators for evaluating a product are two, one is the market capacity, and the other is the degree of competition.

The degree of competition is whether there is a monopoly, what is the degree, how many brands are selling in a certain market, what is the proportion of current existing sales data, and what is the distribution of shelf time. The final hope is that “my new products are available in this market. No chance”.

It took me a lot of thinking time to determine the market capacity.

In my daily work, I am often asked how to determine the market capacity. Often everyone’s common reference standard is to judge under the standard framework of market sales, product quantity, market demand, etc., and I have given suggestions to refine what you need to do. For this product, the total sales of all products that can be made under the same usage scenario and the product you want to make of the same or similar models is the market capacity of your product.

That is: the total sales of mutually replaceable products in the same scenario and the products you want to make is your market capacity. (All major software has reference data)

This method is actually relatively easy to operate, and it can basically solve the main problems of the current work.

But when you think about it, because competing products are changing and the market is changing, past data may only represent the past, and what our products want to do is the future. Due to dynamic reasons, using past sales data as a reference for future market capacity may cause big errors. The reason is that crawler software has a certain delay and error range. The reference is also based on the past market. This is the common pain point of data-based product selection and the main trap of big data selection.

The consequences, either missed product opportunities, or bad sales.

USB Microphone: