Data analysis: knowing a little about the business is equivalent to not knowing anything about it

Many people have always believed that data analysts need to have some understanding of the business in order to perform better analysis. However, there is no specific standard to tell people what it means to truly understand the business. This article will explain it to you, let's take a look!

I have always felt that data analysts need to have some understanding of the business in order to perform better analysis. However, there is a question: to what extent do we need to understand the business?

I have been thinking about this recently, and I think that if you only have a little understanding of business knowledge, it may have a counterproductive effect. Only when you are very familiar with business knowledge can you reach a useful level.

Why do I say so? Here are a few examples.

1. Case 1

Suppose you are analyzing the conversion of different SKUs (different specifications) of a product. If you don’t understand the business at all, you might simply report the conversion rates and final conversion result figures of three different specifications. Although this analysis does not go deep, it has completed your job.

If you have some basic understanding of the business , you will know that too many options will cause confusion to users and increase management costs. Therefore, when you find that the proportion of users who purchase a certain SKU is less than 5% of the total, you will suggest streamlining the SKU and removing the SKU with a very low proportion. Doing so can not only reduce management costs, but also make it easier for users to make decisions.

However, this initial business understanding is not necessarily correct . There is a marketing psychology phenomenon called the bait effect.

What this means is that when people are comparing two similar options, introducing a third option makes one of them more attractive.

For example, a hat costs 59 yuan, and a hat + coat combination costs 299 yuan. At this time, if I add an option to buy a coat alone for 299 yuan, more people will buy the hat + coat. Although almost no one will buy the coat alone, this option makes more people choose the hat + coat combination priced at 299 yuan, thereby increasing sales revenue.

Therefore, if the SKU is streamlined, although the user's decision becomes simpler, sales revenue will also be lost.

2. Case 2

Let me give you another example.

Anyone who studies data analysis must have heard of the beer diaper case. Although this case is fictitious, it is effective for the combination analysis of commodities.

If you know a little about the principles of beer diapers, you can use shopping basket analysis to find out the associations between products. Put related products closer together to increase the probability of them being sold together, thereby increasing sales.

But is this approach correct? If the probability of two products being purchased together is higher, do they have to be put together? You have definitely not thought about this question, because the case study shows that they are put together, and you just need to copy the case study. But in fact, this is not necessarily the case, because this case happened in a supermarket. Supermarkets are fast-moving consumer goods scenarios, and users stay for a short time. This may not be the case in other industries.

If it is in a home furnishing mall like IKEA , when he finds that there is a high correlation between product A and product B, the best strategy may not be to put them together.

Instead, try to place them farther away . Because when shopping at IKEA, users stay for a long time. If you place product A and product B far enough apart, users will need to pass by more booths, so that more products can be exposed and the conversion rate can be increased.

So in this case, if you only know the case of beer diapers, you might apply it mechanically, which may not help the business achieve truly efficient operating results.

what to do

What should I do then?

Although it is useless to have a basic understanding of the business, you still have to understand the business. Learning the business is a direction that will not change. Without quantification, there will be no qualitative change.

The most important thing is not to apply the model, but to think about what is the goal of your analysis?

A large part of the mistakes in the case is due to the lack of purpose in the analysis. People often learn about shopping basket analysis and then do shopping basket analysis. This kind of analysis is not necessarily wrong, but it may not match the problem that the business needs to solve at the time. So the best situation is to wait until there is a specific problem that needs to be solved before conducting the analysis.

For example, IKEA's business wants to analyze how to increase the time users spend in the store or let them see more products.

There is an obvious assumption for this problem, which is to place related products far enough apart. Then shopping basket analysis appears as a tool to solve this problem . Use shopping basket analysis to analyze related products, and then take the opposite action from beer and diapers. But in this analysis process, you will not feel anything abrupt, because your logic is consistent as follows.

• To allow customers to pass through more places
• Find related products
• Place products further away

The original beer diaper example was actually intended to illustrate that data can reveal a lot of hidden information, but its specific use depends on the purpose of the analysis.

3. Summary

The so-called understanding of business through data analysis is a bit like the intelligent driving of a car.

There is no difference between good and bad, only the difference between usable and unusable.

Only when your understanding and analytical abilities reach a critical point can you say that you really understand the business. Otherwise, the little business knowledge you have before this may have a negative effect. Otherwise, it is like driving an assisted driving car and needing someone to watch you all the time, which makes you even more tired.

Some thoughts, I hope they can give you some inspiration.

Author: Jason

Source: WeChat public account "Sanyuan Variance (ID: sanyuanfangcha)"