Article Directory
It's never too late to fix a dead sheep
Due to job changes, I have to say goodbye to the retail scene for the time being. At the beginning, I didn't think too much about it, so I jumped into the "new" retail scene. So far, I haven't done a comprehensive review of the retail scene.
But in fact, there is no need to be harsh on yourself. Back then, as a Xiaobai who had just graduated from school, it was difficult to make a comprehensive judgment without the guidance of a noble person. Most people walk, watch and think. The only difference is that I also hope to sort out the results of my thinking and pass them on to everyone who may need them. The point of view may not be comprehensive or even correct, but it is also the experience of a person who has experienced it. At worst, it can be used as an input for everyone to make judgments, haha.
Ok, let's get to the point.
The retail industry is large
According to the national standard " Retail Business Classification (GB/T 18106-2021)" , which will be implemented in October 2021, retail mainly refers to consumption activities for final consumers (such as residents, etc.). According to whether there is a fixed business place, it can be divided into two categories: retail with stores and retail without stores. Among them, retail with stores can be subdivided into 10 retail formats including convenience stores, supermarkets, discount stores, warehouse membership stores, department stores, shopping centers, specialty stores, brand stores, collection stores and unattended stores; retail without stores Including online retail, TV/broadcast retail, postal retail, unmanned vending equipment retail, direct sales, telephone retail, mobile stall retail and other 7 retail formats.
Next, let’s intuitively feel the scale of the retail industry through a set of data. Here, we assume that online retailing is equivalent to online retailing, and social consumer goods retailing is equivalent to retail market (because there is no direct data on online retailing and retail market).
| years | Online retail sales (trillions) | Total retail sales of social consumer goods (trillion yuan) | Proportion of online retail sales | GDP (trillions) | Retail/GDP |
|---|---|---|---|---|---|
| 2017 | 7.2 | 35 | 0.21 | 83 | 0.42 |
| 2018 | 9.0 | 38 | 0.24 | 92 | 0.41 |
| 2019 | 10.6 | 41 | 0.26 | 99 | 0.41 |
| 2020 | 11.8 | 39 | 0.30 | 102 | 0.38 |
| 2021 | 13.1 | 44 | 0.30 | 114 | 0.39 |
| 2022 | 13.8 | 44 | 0.31 | 121 | 0.36 |
From the above table, at least three conclusions can be drawn:
(1) Retail accounts for about 40% of GDP. This is a very large proportion. Therefore, this matter itself is indeed very important, and it is directly related to the national economy and the people's livelihood.
(2) The absolute value of online retail sales in 2022 is 13.8 trillion. This value may still not be intuitive, let's compare it a little bit. Wal-Mart is a representative of small profits but high sales. Its total sales in 2022 will be 572.754 trillion US dollars, with a net profit of 13.676 billion US dollars and a profit margin of 2.4%. Even if we reduce the profit margin to 1%, then 13.8 trillion sales can bring 100 billion+ net profit. JD.com is No. 1 in the " 2022 China Online Retail TOP100 ", with annual sales of 800 billion+ yuan. If it is converted into profit according to the 1% standard, it is expected to reach 8 billion+ yuan. So this scenario is very attractive to the company.
(3) Online retail sales account for about 30% of the total retail sales, and the growth rate is gradually approaching zero. This is not consistent with our intuition. Internet + is in full swing. After more than ten years of transformation of the retail industry by domestic Internet companies, the proportion of the Internet has changed to 30%. According to the " 51st Statistical Report on Internet Development in China " released by the China Internet Network Information Center , as of December 2022, the number of Internet users in China is 1.067 billion, and the number of online shopping users is 845 million. Therefore, the population base is already very large, and it is very difficult to increase the proportion of online shopping. There is no growth in the market, and companies can only "coil" for their own development.
Reaching new heights in retail
Since you want to roll, you must first know which directions retail can roll, and then you must clarify the history of retail development. Here I recommend Liu Run's "New Retail: The Road to Low-Price and Efficient Data Empowerment". This section is mainly based on the content of the book to sort out.
Retail is defined as a "field" that connects "people" and "goods". The earliest market was a bazaar; then department stores and supermarket chains were added; and now an e-commerce platform has been added. With the evolution of the field, the types of "goods" that can be connected have increased, and the scale of "people" has also increased.
Next, let’s understand retail from the perspective of “people”, “goods” and “market”.
People: traffic x conversion rate x customer unit price x repurchase rate people: traffic\times conversion rate\times customer unit price\times repurchase ratehuman traffic×Conversion rate×Customer price×Repurchase rate
This is relatively easy to understand, so I won’t go into details.
Goods: D − M − S − B − b − C Goods: DMSBbCGoods: D−M−S−B−b−C
D=Design, design; M=Manufacture, manufacturer; S=Supply Chain, supply chain; B=Business, large shopping mall; b=business, small store; C=consumer, consumer.
The above is a complete chain of a product from design, production to consumer market.
Field: information flow + capital flow + logistics field: information flow + capital flow + logisticsfield: information flow+Cash flow+This is a new definition of logistics
. Let me give you an example. When we go to the supermarket to buy a product, we will touch the quality of the handle to see if it has expired, etc. This is "information flow"; if it feels good, put it in the shopping cart , Push it to the cash register to pay, this is "fund flow"; after purchasing, drive yourself or take the supermarket shuttle bus home, this is "logistics".
With the above basic knowledge, let’s take a look at the new ways of retailing so far.
For ease of understanding, here is an example: Costco. Costco is the second largest retailer in the world. In the 2017 Fortune 500 list, Costco ranked 16th. In terms of short-circuit economy, Costco purchases directly from the manufacturer (M) and displays them in its own store (B), which short-circuits the supply chain (S) in the middle and improves the efficiency of the chain, which belongs to M2B; in terms of data empowerment, Costco will use big data to select products that it thinks have the potential to be "explosive" on the shelves, and the packaging is large and the quantity is sufficient, which can bring an excellent experience to consumers; in terms of floor efficiency, the conversion rate has been improved through the membership system The unit price per customer, low price and high quality can promote the repurchase rate, so that its per-square-foot efficiency can reach twice that of Wal-Mart.
Costco represents a membership-based retailing approach. Its core profit model is not the profit brought by the product itself, but the income from membership fees. This model has basically worked, and there are similar companies like Wal-Mart's Sam.
The other two retail models that are relatively popular in China are: instant retail and community group buying. Instant retailing focuses on half-hour/one-hour delivery. At present, some well-known companies include Hema Xiansheng, Dingdong Maicai, Meituan Maicai, and Pupu Supermarket. This model has not yet fully worked out, and each company has suffered losses to varying degrees. Liangliang’s Daily Fresh is an example.
The soul of community group buying is affordable. Around 2020, a large number of players entered the community group buying, such as Xingshengyou, Shihuituan, Jingxi Pinpin, Taocaicai, Meituanyouyou and Duoduomaicai, etc. However, this model burns money too fast, and has not found a way to make a profit so far. At present, only Meituan Youyou and Duoduomaicai still account for a large share of the market.
Practical experience in operational optimization
In view of the large scale of the retail industry and the continuous exploration of "new" retail, there is room for algorithms related to operational research optimization. If you look at the big picture of retail, it is actually easy to find that the place where the operation research optimization algorithm can play is mainly the S module in the short-circuit economy, that is, supply chain optimization.
From the perspective of problem scenarios, logistics optimization can be applied to store location/planning to increase operating income; it can be applied to personnel scheduling/setting to improve human efficiency; it can be applied to real-time/timing scheduling of manpower/resources to improve automation efficiency and optimality.
From the perspective of technology stack usage frequency, the two most widely used algorithms are integer programming and heuristic algorithms, and gradient algorithms are less involved.
From the perspective of related technologies, the machine learning algorithm has the highest degree of integration. In many places, the predicted results are required as one of the inputs of the operational research optimization algorithm model.
From the perspective of practical difficulty, compared with the best modeling design, the adoption rate of the algorithm after landing is often a greater challenge.
The main reason for this phenomenon is that the current understanding is that the problem itself may indeed be relatively complicated, but after adding more realistic constraints, the problem is reduced to a relatively small-scale problem, and the algorithm designed by the business side based on its own experience The effect is already very good, and the additional benefits that the algorithm can bring are small, and even reduce the flexibility of the business side.
From the perspective of value manifestation, it has improved the business, but its upper limit is likely to be limited by the specific operating model and process; the improvement point is mainly manifested in two aspects: process automation and better result indicators.