The sales data analysis ideas you must sort out before "Double 11"

The case is this:

An e-commerce company that sells gifts on an information flow platform had lukewarm sales for the previous 10 months, with ups and downs. Monthly sales had been hovering around 400,000 yuan, but last month's sales suddenly exceeded 1 million yuan. Moreover, this sales volume was achieved with advertising expenses of less than 100,000 yuan. This made the boss very excited, as if spring had come! The torment of the past 10 months is finally coming to an end!

I couldn't help but pour cold water on it:

Does sales exceeding expectations mean that operations are on the right track?

If not, how can we discover the business insights hidden behind the 1 million yuan figure?

Through analysis.

When we are faced with data, we cannot just look at the data itself or talk about the data alone, but we must use a complete data analysis framework to interpret the data.

There is a classic 6-word strategy for data analysis:

Segmentation, comparison, and tracing.

Specifically:

● First break down the problem from different dimensions;

● Then build a reference system based on the data in each dimension for comparison to find the weak links or parts of the business;

● Finally, analyze the causes and find improvement plans.

Only in this way can we discover the real business insights behind the 1 million yuan and better formulate the next strategy.

01 Segmentation

Segmentation means finding the influencing factors of an indicator that needs to be analyzed by breaking down different dimensions.

Dimension originated from geometry and refers to the path connecting two similar spaces. When used in data analysis, it mostly refers to a perspective. For example, gender, region, time, distance, etc. are all perspectives often used in data analysis.

For example, if we want to use time as a dimension, one method is to understand the development trend of things by comparing the previous and subsequent times in the same space. For example, in this case, the sales volume last month was 400,000 yuan, and this month it was 1 million yuan. We can say that the month-on- month growth rate of sales volume is 150% . This is a comparison in time, also known as a longitudinal comparison.

You can also understand your current situation and gaps through horizontal comparison in space and time, such as comparison of population and GDP of different countries, comparison of income and number of users of different provinces, comparison between different companies and departments. These are comparisons between units of the same level, referred to as horizontal comparisons.

The premise of disassembly is to find the appropriate dimension, so how to find the dimension?

The essence of finding segmentation dimensions is to discover the influencing factors of the dependent variable. For example, in this case, the dependent variable is sales.

There may be many factors that cause the total sales volume to change from 600,000 yuan to 1 million yuan. We can use 6W2H or the marketing "4P" framework to analyze, list all the dimensions that may need to be considered, and then based on the specific circumstances of the case, see which dimensions it will involve, and select the appropriate dimensions for the next step of decomposition and analysis.

Here we use 6W2H to demonstrate the process to you:

1. What, what product do you sell?

It can be expanded according to specific issues, such as:

What is your product line?

What are your competitors?

What are your substitutes and complements?

What are the inherent choices of your target consumers?

…

Suppose we now need to break it down from the product line dimension. Generally, a gift e-commerce company may sell more than one kind of gift. It may also sell high-end signing pens, telescopes, etc. Then the question that can be broken down based on this is: What is the sales contribution rate of each product?

According to the 28 principle, the ideal situation is that 80% of sales come from 20% of products. We need to know what the hot-selling products are and what the products that are not selling well are? Only with clear segmentation can we proceed to the next step of analysis to understand the reasons for sales changes and judge future trends.

2. Who, that is, consider who is selling?

For example, who within the company is responsible for selling this product?

How many peers are there in the external market selling the same product?

If we consider the internal aspects, we can break down the problem into which departments are selling? Which project managers are responsible for sales? How do they each contribute to sales?

If you consider the external market, you can analyze what the sales of your peers are? How does this 1 million yuan level rank among its peers?

3.Whom, which means who are you selling to?

Who are your customers?

What are the types of customers?

Who are the users? Who are the influencers? Who are the decision makers?

What is the consumption level?

What is the purchase frequency?

…

In this case, we can think about how many types of people there are who buy gifts online? What purchasing characteristics do they each have, and so on.

4. When, when to buy?

What is the pattern in the time distribution of customer purchases?

For example, in this case, can we compare and see if there is a pattern of cyclical changes in sales?

5. Where: Where do customers buy?

We can take a look at what sales channels are available, such as online or offline? If everything is online, then is it an information flow platform or an online shopping mall? etc.

In this case, the boss has set up several channels or platforms to sell his products. Among so many channels, which channel or channels have better sales?

6. Why, that is, what is the customer’s purchasing motivation?

Why should customers buy my product?

What are the customer consumption scenarios?

What are the customer pain points?

What environmental factors influence customers’ purchases?

etc.

For example, in this case, we can think about whether the customer’s purchasing motivation is more for gifts or for personal use? What motivates them to buy gifts? etc.

7. How: How do customers buy?

How to buy can be a study of payment methods, such as cash on delivery or payment first and then delivery? Is it WeChat payment or Alipay payment? etc.

How to buy can also be how the elements of the marketing mix influence consumers' purchasing decisions. What factors do consumers value in purchasing decisions? Is it the convenience of purchase? Or the design of the product? Or the price?

8. How much , that is, how much are customers willing to pay for the product?

This involves the pricing strategy . Should we adopt low-price penetration? What about skimming at high prices? Or discriminatory pricing? etc.

You can also examine how the customer's purchase cost is composed, whether it includes understanding cost, monetary cost, risk cost, time cost, convenience cost, etc., and analyze specific issues specifically.

In this case, what needs to be considered at this point is whether the current pricing strategy will have an impact on sales? Which pricing method is more conducive to increasing sales?

02 Contrast

After completing the segmentation in the previous step, we need to start comparing data of the same dimension to understand the current status of the business and find the weak links or parts of the business.

Comparison is mainly done by establishing a reference system. There are a few points to note here:

1. Find out who is competing with whom

The meaning of “who” here is relatively broad, for example, are they from the same unit competing? For example, the same sales amount, 3,000 RMB and 900 Euros cannot be directly compared. You have to unify the units first and then compare.

2. Figure out how to compare

That is to say, the objects of comparison should be comparable. For example, the sales volume of a company’s product cannot be compared with the total social retail sales of the product in the province. They are not in the same order of magnitude, so how can they be compared? Comparisons can only be made with those of the same magnitude.

3. Find out what to do after the competition

If all the steps are accurate, then after the comparison, we will be able to know our approximate state. If this state is relatively good, we will continue to maintain it. If the state is relatively poor, we can start to improve and optimize it.

For optimizers, weak parts and weak links are the entry points for optimization work. In practice, good optimizers are often very sensitive to "differences".

Suppose we find through a comparison of product dimensions that among the 10 categories of gifts sold by the company in this case, the comparative sales of most products are quite good, and all steps in the operation are also performing well, and some have even reached the limit. Only one watch sells particularly poorly, ranking last in sales. Last month's sales were only 3,000 yuan, accounting for only 0.3% of the total sales - that is, this watch contributed little to the 1 million sales last month, and is a weak link in the company's sales.

In this case, should we look for the best performers to optimize and upgrade and challenge higher limits? Or should we try to make up for the shortcomings first?

Here we give a reference direction for judging the trade-offs:

● It can be judged based on the optimized input-output ratio;

● You can choose based on the difficulty of optimization and the room for improvement;

● Focus on the company’s resources;

● Refer to the company's business objectives and business strategies;

● Consideration of other influencing factors, such as current market policies, industry development trends, peer competition, etc.

Assuming that based on the analysis of the above factors, we conclude that the sales entity company in this case is very optimistic about the development prospects of the watch category, and the company's future business strategy is multi-pronged development, then we can believe that increasing watch sales is a top priority, and the next work is to use increasing watch sales as a starting point to increase overall sales.

03 Tracing the source

Generally speaking, when a product has poor sales, what is the first reaction of the people involved? Blame!

Blame the optimizer for not placing the ad properly? Blame the product manager for choosing the wrong product? Or is it because the price is too high? All kinds of guesses, all kinds of blame, and all kinds of shirking of responsibility.

Here I would like to remind you that unfounded speculation will not help . An effective solution is not to guess which one it is, but to use a model to trace back all possible problems along the way and find the source of the problem.

We use the fishbone diagram to analyze the operational problems that e-commerce companies may encounter when launching information flow advertising .

(Note: In order to view the following figure more completely,

It is recommended to turn the phone screen sideways after clicking the picture)

According to the diagram, we can understand that the entire e-commerce delivery process involves six aspects, namely the initial selection of models , followed by targeted pricing , followed by delivery-related targeting, creativity, and landing pages , and finally customer service, logistics and other back-end aspects . Each step here can be broken down into many small steps.

For example, the selection process.

Product selection for e-commerce is a very complicated process, which requires high marketing capabilities and market sensitivity of the selector. So usually we can consider the following when choosing a model:

1. What is the key to competition in the industry in which your product operates?

For example, the main competitive key in most fast-moving consumer goods industries is brand. If the product we want to choose belongs to this field, then we have to consider whether the brand of this product has appeal.

For example, if you are selling mobile phones online, under the same conditions, the results of selling iPhone and "YOTA" phone will be different.

2. Is the product marketable?

There is a difference between selecting products online and offline. Products that require actual experience or professional consulting services to complete a transaction are not very suitable for direct online sales. Products with too high a conversion threshold will also have a higher user loss rate when sold online.

For example, the sales of some insurance products require dedicated services to complete the transaction. The threshold for online sales conversion is too high, and it is reasonable that the conversion cost is high.

There are many other issues to consider, such as product profit margins, market competition, entry barriers , etc. I will not elaborate on them here. We can analyze them in detail when we encounter specific problems.

In this case, the company's selection was mainly based on the data of physical gift retail stores in the company's location. Judging from the sales of these retail stores, this watch is a hit, with more than 40% of customers repurchasing it as a gift within three months!

What does 40% mean? Generally speaking, a 20% repurchase rate for offline gift sales is already good, and 40% is equivalent to doubling it. It is precisely because of this that the boss of this e-commerce company has great confidence and thinks he can give it a try by selling online.

Now let’s talk about the customer service aspect.

The customer service link mainly considers the user experience . Generally, the factors that affect the user experience are as follows:

1. Is the response fast?

For example, if you asked customer service a question and no one responded for a long time, how do you think the user would feel?

According to relevant data, if the response time exceeds 15 seconds, the consultation may fail. The more urgent the consultation, or the more precise the user's needs, the stronger their desire for quick feedback, and they may not be able to wait even 15 seconds.

2. Is the question-to-answer ratio high?

This refers to the interaction between customer service and users. Are they enthusiastic and welcoming? The user asks you three questions and you answer one. If your question can explain the problem, it's fine. If not, the user experience will be bad, and naturally, the conversion rate will not be very good (some special industries, such as reverse marketing in the luxury retail field, are not within the scope of our discussion today).

3. Is the speech professional?

Let us first explain what professionalism is: one refers to whether the customer service staff ’s online sales level is up to standard, specifically referring to the customer service staff’s communication skills and ability to force sales.

The other refers to whether the customer service staff has professional knowledge of the product field. Whether or not you can answer users’ questions accurately and effectively will directly or indirectly affect their level of trust in the product. For example, in certain specific product areas, such as myopia glasses, the customer service staff’s mastery of technology and professional level will directly or indirectly affect the quality of his reception, and thus affect the user’s willingness to buy.

In this case, the customer service team of this e-commerce company is shared with the customer service of other products. From the response time to the question-answer ratio to the customer service's order-taking process and the professional level of order-taking language, they all meet the requirements of marketing professionals. Moreover, the actual level of the reception staff is not much different from that of the store clerks, or even slightly better. Therefore, we can roughly rule out the impact of the customer service link on sales.

Suppose, through the fishbone diagram, we analyze the marketing links layer by layer and find that the link most likely to have problems is pricing, because the current pricing is a bit too low compared with peers. Then, will customers think that cheap products are not good because the price is too low? Insecure about quality? So you don't want to buy it? ——As I said, guessing is not rigorous, and we need to use professional tools for analysis.

Here we recommend the 5 why analysis method.

The 5why analysis method originated from Toyota in Japan. It is an analysis method commonly used by Toyota's former vice president Taiichi Ohno when he discovered a problem: by continuously asking "why?" to find the real cause and solution of the problem, it is applicable to many fields.

Taking the case given in this article as an example, we can explore why watch sales are low by asking the following questions.

Questions begin:

1. Question: Why hasn’t the watch been converted even though the price is so low?

1 Answer: I’m afraid it’s fake.

2Q: Why are you worried that it’s fake?

2Answer: Consumers have an inherent perception that products with lower prices than those of other products may be fake.

Question 3: Why are products with low prices considered fake?

3Answer: Price is a reflection of value. High price = good quality, cheap price = poor quality.

Question 4: Why do users think that high price = good quality, and cheap price = poor quality?

4 Answer: When people judge an event that is complex, ambiguous and uncertain, if there is no effective method, they will often take some shortcuts to simplify their thinking. For example, when judging the relationship between product quality and price, there is a lack of decision-making basis. Users may determine the anchor value based on past experience. For example, if they have bought something very cheap before, but it turned out to be of poor quality, they will think that cheap price = bad product.

Question 5: Why do people rely on simplistic mental shortcuts to make decisions?

Answer 5: From Thinking, Fast and Slow, written by Daniel Kahneman, the 2002 Nobel Prize winner in Economics, we know that there are two systems in the brain. System 1 operates unconsciously and quickly, and does not require much brainpower; System 2 shifts attention to brain activities that require mental effort, such as complex calculations.

The brain also likes to minimize the cost of decision making.

In daily life, it is enough to use only System 1. System 2 will be triggered only when System 1 encounters a problem that it cannot solve independently, such as an operation like “17×24”.

The problem is that during this process, System 1 will often come to disrupt us. The most troublesome thing is that we are unable to shut down System 1 which likes to take shortcuts. It can be said that it is very difficult to get System 2 to perform its work well.

(Note: 5why does not mean that you can only ask 5 “whys”. You can ask more or less. When you feel that you have found the root cause and solution of the problem, you can stop. For example, we can continue to ask the above 5 questions. In the end, you will find that this is actually a question about human evolution...)

So the answer is:

The selling point of cheapness that you originally thought was not recognized by the human brain. The cheapness itself has not reached the level to trigger the rational system 2. In other words, users cannot rely on the rational system 2 to analyze that your price is actually super cost-effective and give instructions to buy it.

On the contrary, System 1, which likes to intervene in various issues, directly sends the signal that cheap products are not good, hindering users' actions and ultimately leading to such low sales.

Then the corresponding solution is:

Don't fight against the decision-making mechanism of the human brain, and don't try to bypass System 1 that likes to interfere in various issues. Since people think that cheap products are not good, you don't want cheap things.

So next we can optimize the pricing strategy, and then go back online to test whether the sales situation is consistent with expectations, and then make further optimizations and improvements, and repeat this cycle for continuous improvement.

The entire analysis ends here.

Let’s go back and look at the case mentioned at the beginning of the article. The data of 1 million yuan itself actually has no meaning. Only by substituting it into a specific business context and through comparison, segmentation and tracing can we interpret the business value behind it.

Of course, the problems encountered in actual operations will definitely be more complicated than the cases we talked about, and there will be many more factors to consider, but the basic ideas for solving problems are the same.

After all, data analysis is part of business, and all analytical work is to help us achieve our business goals.