When writing a data analysis report, don’t make these mistakes

In the field of data analysis, an accurate, clear and insightful report is a key support for decision-making. However, when writing data analysis reports, many people often make mistakes in data accuracy, clarity of expression, correctness of direction and rationality of judgment. These mistakes may mislead decision makers and even damage the professional reputation of data analysts.

When writing a report, I am most afraid of hearing these three words. Especially now, many students are intensively writing annual reports. Today, I will show you how to avoid the four major reporting errors and how to deal with the four common challenges. Students, please remember to click "like" first, and you can read it slowly when you have time.

1. Four common mistakes

▌Error 1: Data error.

for example:

l The unit was omitted: sales volume was 18 million, but it was written as 180.

l The numbers were miscalculated: the number of users was 15,000 but it was counted as 16,000.

l Indicator used incorrectly: The number of consumptions was 1500, but the number of consumers was 1500.

This kind of mistake is very low-level and very serious! Because correct data is the basis of analysis, data errors will damage the trust of leaders in data analysts, so they must not be made.

▌Error 2: Incorrect expression.

for example:

This kind of error is: the wording is not good, the number is correct, but the person who reads the number will misunderstand it. The best way is: after writing it, let someone who is not familiar with the report take a look, and quickly point out the problem.

▌Error three: Wrong direction.

This kind of mistake is often caused by students who work on data not understanding the business, and just talking about numbers, and talking about nonsense that has no practical significance. To change this problem, you can only practice more and strengthen your understanding of the business.

▌Error 4: Misjudgment.

for example

It looks funny, but the actual situation is so embarrassing.

For example, we immediately ask three questions:

What is your company’s strategic goal this year?

l What are your user numbers, sales, and profit targets for this year?

l What are the user numbers, sales volume, and profit targets that should be achieved in April?

Nine out of ten people couldn't answer it. The one who answered it happened to be the student who was working on the daily and monthly reports today. Of course, this is not all the fault of the data. In many companies, the business and data are not communicated well, and even the people in the business department are confused, not to mention explaining it clearly to others. The most typical example is as follows:

Misjudgment is the most difficult of the four errors. Because the first three are basically the data's own business, as long as you practice hard, you can improve, but misjudgment is not something that the data can handle on its own. If the business does not cooperate, or even deliberately muddies the waters and calls a black cat a white cat, it is very likely that the data will be unable to explain itself. So we have to look at it separately and see how to deal with it.

2. Avoid incorrect operation steps

▌Step 1: Distinguish the report type.

Report type: first report or nth report

Report content: report results/evaluate the quality/analyze the causes/predict trends

Report background: Business side has prediction/no prediction

This step must be done before you pick up the pen, to recognize the situation and not fight an unprepared battle.

If this is your first report or notification of results, just state the data in a straightforward manner.

If you want to evaluate whether something is good or bad, you must communicate the evaluation criteria with the business party in advance.

If you are analyzing the causes, you need to combine it with the business side's predictions and give priority to the issues that the business side is concerned about.

If you are predicting a trend, you must clearly explain the basis for the prediction.

▌Step 2: Clarify business goals.

Clarifying the goal is to establish the basis for judgment standards. As long as the goal is clear, the standard will not be far behind. If the goal is a single indicator or result data such as performance, income, or number of users, you can directly use the decomposition method to find the judgment standard at the current time.

Special note: If the operation wants to talk to you about "our goal is to increase the natural growth rate by 5%", be especially careful! Because the natural growth rate is not a directly collected indicator, and it is not even a fixed indicator. There are many algorithms, each of which has certain bugs, and there is no absolutely good natural growth rate algorithm (as shown in the figure below).

So everyone should remember: if it involves judgment, you must communicate in advance, in advance, in advance (important things should be said three times). For example, to evaluate the effectiveness of next month's activities. Ask the business in advance: What goals do you want to achieve? What is the standard for judging good or bad? It is best to let the business give the expectations for next month in advance! Predict whether it will rise or fall, and tell it in advance, this can save 90% of the trouble.

▌Step 3: Verify data and check expression.

These are all basic skills, and they are related to specific data collection habits. If there is a chance, I will find a data set to talk to you about it. I won’t expand on this here. But no matter how you look at it, you will always miss some bugs. The best way is to have an acquaintance on the business side, who can take a look at the report in advance. The stranger, the more likely they are to see the problem at a glance.

▌Step 4: Face challenges calmly.

It is inevitable to face challenges. Don’t rush to argue or change things. First, make it clear:

Is there any problem with the data itself?

Is there any problem with the description of the data?

Is there any problem with the analysis dimension?

l Whether the judgment standard has been changed

There is something wrong with the data/the description is unclear. This is our own fault. Stand up, take the blame, admit the mistake, and correct it!

The business does not recognize the analytical dimension. On the surface, it seems to be a communication problem, but in essence, it is a lack of understanding of the business. This is a purely technical issue that can be improved through practice.

If the judgment criteria are proposed by us and the business thinks they are not suitable, then we will redo them according to the business's proposal. This is also a purely technical issue. As long as everyone agrees on the criteria, it will be fine.

If the judgment criteria were proposed by the business, but they went back on their word! This is very likely because the business found that they did not do well and wanted to cover up the problem. At this time, you have to study how to avoid being blamed by the business. To make a good report, you need not only methods, but also communication skills.