User portraits are criticized as "useless"! How can we make them useful?

As an important part of data analysis reports, the value and practicality of user portraits are often questioned. When writing user portrait reports, many data analysts may face the problem that the content of the report is too basic or lacks depth, which leads the business side to think that the report is "useless". This article will explore in depth how to improve the practicality and value of user portrait reports, including avoiding common pitfalls, establishing effective judgment criteria, and proposing practical business suggestions.

Today I will continue to share how to write a data analysis report. Many students asked: "How to write a report on static data?" This is especially true for user portrait reports. Many times, the business asked: "Look at our user portraits." As a result, the report was criticized: "I know everything, what's the point of writing this?" How to do it? Today I will share the system.

1 Useless reports look like this

When it comes to user portraits, many students' reports look like this:

● Male to female ratio 4:6

● 40% are over 30 years old

● Average annual consumption: 500 yuan

● Active users for more than 1 month: 55%

● ……

Often, such reports are very hard to write and require a lot of data. In the end, you will get a response like, "I already knew that. Can you give me something more meaningful?"

Some people think that it is not enough to just provide descriptive data, so a model must be built. The most straightforward method that comes to mind is RFM, so the three dimensions of data, R, F, and M, are extracted, each divided into five segments, 5*5*5 divided into 125 categories, and then clustered into 5 major categories and 125 minor categories, each of which is carefully labeled: "This user bought 500 yuan once, and has not sold for 5 days, so we have to let him buy!"

In the end, I received a barrage of questions:

● If you say he will buy it, then he will buy it!

● What to buy! Where to buy!

● How to get him to buy

● How to inform him to buy!

● So what if you don’t buy it!

● So what if I bought it!

● If only he would have bought it!

...I was so dizzy from the explosion...

What exactly is the problem?

2 Core reasons why reports are useless

The core reason is just one: there is no standard for judgment.

● Is the gender ratio a key issue?

● Is the 4:6 male-female ratio a problem?

● So what if the male to female ratio is 3:7

● There is no standard for judgment

It makes people feel confused.

What’s interesting is that monitoring data has its own standards. When we look at a curve, even if we don’t define a “target value”, the change in the curve itself can become a criterion for judgment.

For positive indicators such as sales, profits, and number of users (the more the better), growth itself is good, the faster the growth rate, the better, and the larger the absolute number, the better.

For negative indicators such as costs, risk losses, and complaints, a decline in itself is a good thing. The faster the decline, the better, and the smaller the absolute number, the better.

Of course, it is rash to judge like this, and there is a possibility of large ups and downs, but it is okay to take a rough look.

But most static data has no standard. For example, is a 4:6 ratio of male to female necessarily good or bad? Not necessarily. Therefore, if you cannot form a judgment when you see this kind of data, it is difficult to generate value.

Moreover, the business department itself has a certain feeling about the user situation. For example, "the users we serve are mainly females." Even if they don't see 4:6, they know that there are more females. Seeing 6 and feeling similar, they will naturally say, "I knew it a long time ago."

As for the judgment that "because the user spent 500 yuan once, he should spend 600 yuan", it is just a matter of numbers and is completely unreasonable. Numbers are just a record, and behind the numbers is a real and vivid scene. It is meaningless to talk about numbers without considering the business scenario.

Just like when you go to eat noodles, you enter the noodle shop and shout, "Boss, I want a bowl of beef noodles." The boss yells, "No! You don't want one bowl! Our senior big data analyst from artificial intelligence says you need two bowls! Give them to me!"

How do you feel at this time?

So there is only one key to breaking the deadlock: find a standard. Let the data express meaning, not just a string of numbers.

3 Ways to improve the usefulness of reports

First, you need to distinguish the scenarios. Note that if it is the first report, you can start with a fierce complaint. It may be because of a new leader or because the CRM system has just been implemented. In short, the business side has no understanding of the user situation before. At this time, it is very helpful to list a bunch of data in detail. It can let everyone understand the situation in detail and establish cognition.

However, starting from the second report, we can no longer be so straightforward; nor can we just throw plain words at the business parties who already understand the situation. We need to find the judgment criteria. There are three common ways to find them:

1. Find the standard from the problem. As shown in the figure below

▌

2. Find the standard from the goal. As shown in the figure below

3. Find standards from business. As shown in the figure below

Of course, all three methods require the following three points:

● There is sufficient communication between data and business

● Understand the business background (work objectives, design ideas, implementation plans)

● Understand the basic logic of business operation and corresponding data tables

But in many companies, these three conditions are not met.

It may be because the business department thinks they are God, and think they know everything, and they just need a guy to run numbers, so they recruit SQL writers instead of data analysts.

It may also be because the company treats the data analyst as God, thinking that as long as he does the data analysis work, he is omniscient and omnipotent. Other people in the company don't say anything, and the data analyst knows everything just by a flash of golden light...

In short, if the data report is useless and the data is out of touch with the business, it can only mean that the company is not doing well. Both the business and the data are responsible. Instead of passing the buck to each other, it is better to communicate honestly to achieve better results.

However, some students said: Teacher, it is obviously the same report, why do some people say it is well done, while others criticize it for being low-level? It feels so difficult. That is because the speaker is unintentional, but the listener is intentional. The only person who can make the whole country happy is Zhao Benshan. If you replace him with Guo Degang, some people will think he is vulgar, so you must eat according to the food.