Help! How to write the recommendations section of a data analysis report?

When writing a data analysis report, how to convert data into valuable suggestions is a difficult problem faced by many data analysts. This article uses a specific attendance problem scenario to detail how to start from the problem, verify assumptions, sort out logic trees, and support data, and finally draw reasonable conclusions and suggestions.

Don’t just report numbers!

Make actionable suggestions!

Many students who work with data have been yelled at by their leaders and colleagues. However, how can we do it? Teacher Chen has a classic exercise that I would like to share with you today.

Problem scenario: In the company, a student's attendance sheet is as follows:

The student: "I worked overtime, it's normal to be late!" "Everyone is late, why are you catching me alone?" "It suddenly rained, there's nothing I can do!" "It's just all kinds of accidents!"

Leader: "I think it's your attitude that's the problem! Don't bring up other things!"

The two of them were arguing fiercely! How should we analyze this?

1. Suggestions that cannot be implemented

Are there any classmates who wrote like this?

• There are 22 working days in this month, 11 working days are late, and the lateness rate is 50%.
• The second week was the most late, with 4 days of lateness and a lateness rate of 80%.
• The third week had the least number of late arrivals, with 1 late day and a lateness rate of 20%.
• Too many days late, it is recommended to lower the number
• Monday is too late, it is recommended not to be late on Monday

Isn’t it true that many people’s reports at work look like this? Obviously, such reports are not up to standard! This is just reading the chart again, without responding to the issues that people really care about. “We must lower the price” is also an empty talk.

In the dispute, the core business question is: Is it really an accident, excusable or an attitude problem? Only by addressing this issue head-on can we get a satisfactory answer. Obviously, the data in front of us is not enough. We must first list the ideas for verifying the hypothesis, and then add data to fully answer the problem.

2. Ideas for solving the problem

The basic order of solving problems is to make assumptions → find evidence → verify the truth → draw conclusions. Using data to analyze business problems must be a process of gradual verification from rough to fine.

For example, a simple sentence like "My home is far away, so I am easily late." We can extract a hypothesis: Is my home really far from the company? To verify the hypothesis, we don't need complex data. Just enter the starting point/point in Amap and you will see:

• How far away
• How long does it take to take the subway?
• How much does the ride cost and how long does it take?

Some simple analysis conclusions are ready to come out (as shown below)

Solving problems with data is so refreshing. No need to argue, just speak based on facts.

Of course, when there are many assumptions, we can have a certain order of verification. For example, we can first list the "extenuating circumstances" assumptions, such as:

Assumption 1: Working overtime the day before

Assumption 2: Everyone is late that day

Assumption 3: There is extremely bad weather on that day

After listing the hypotheses, find evidence for each one.

Assumption 1: Overtime work the previous day → Select the time of work/leave work the previous day to see if overtime work was done

Assumption 2: Everyone is late now → Take the clock-in records of everyone on that day and see the percentage of latecomers

Assumption 3: There is extremely bad weather on that day → Check whether there is heavy rain or snow on that day

If there is no data evidence, it means that the hypothesis was made up and can be directly overturned. If it is found that the real situation is consistent, for example, the person really worked a lot of overtime and the missing data is "excusable", then the conclusion can be output: "It is true that they worked too much overtime, and it is recommended to remind them", which clears the person's name.

3. From Simple to Deep

Sometimes, simply verifying a few assumptions is not enough to fully explain the problem. For example, working overtime the day before could be due to personal incompetence/slacking off, or it could be due to the large number of overall tasks. Therefore, further detailed assumptions can be made:

Assumption 1.1: The workload of the entire department is heavy

Assumption 1.2: The department's workload is not large enough, but individuals are burdened with too much

Assumption 1.3: Individuals do not take on much responsibility, but their abilities are poor and they work slowly

Further verify the three subdivision hypotheses, so that the analysis can be gradually deepened from coarse to fine. Finally, a complete answer idea is formed. After reading this, students can readjust their answers and read the subsequent explanations.

4. Overall thinking

By combining multiple analysis assumptions in order, we can get the analysis logic tree shown in the figure below. As you can see, this analysis logic tree starts from the "influence of overtime work" and prioritizes excluding objective influences such as collective overtime and too much work assigned. This order can effectively prevent employees from being wronged (as shown in the figure below)

After listing the logic tree, you only need to substitute the relevant data to find the key points and distinguish the truth from the false.

For example, if you find that out of 11 latecomers, 8 of them were due to too much work, and their workload was significantly higher than that of their colleagues in the group, then it is confirmed that they were wronged. If you find that out of 11 latecomers, only 2 were actually working overtime, and the other 9 were not working overtime and did not take a taxi, then it means that there may be a real problem with their attitude.

Data guides us to find answers that are closer to the truth.

With the above preparation, the derivation suggestions can be justified and very specific (as shown below):

5. Back to the real work

Of course, the above is just a simple example, but it clearly reflects the problem in reality:

Business departments often think in their own way and like to say: "This is a problem of the overall environment", "This is an unexpected problem", "I have tried my best"

Data departments often get caught up in the numbers game, focusing too much on calculating year-on-year and month-on-month changes, and fail to make assumptions, find evidence for assumptions, or refine assumptions.

This is not conducive to drawing correct conclusions and suggestions.