How data analysis drives scientific decision-making

In today's era of information explosion, data analysis has become an important support tool for management decision-making. However, how to reasonably use data analysis methods and techniques and effectively apply them to business decision-making is a challenge faced by every data analyst. This article will introduce some methods and techniques to help data analysts better support management in making scientific and accurate decisions.

“Data analysis should support management in making scientific and accurate decisions”

——This is the requirement of many companies for data analysts. However, the question is: how to support them? Why do you work hard to code a lot of data, but still be told: it is useless?

1. Demystifying decision making

Many students get scared when they hear the word "decision-making". The first impression of "decision-making" is all kinds of high-sounding words that are hard to understand, such as "seize opportunities", "work hard", "get down to earth"... How do these things have anything to do with data?

If you only look at these fantasy words, they really have nothing to do with data! So if you want data to support decision-making, the first step is to strip away the mystery of decision-making and use the simplest and most straightforward data model to describe the decision-making process, so that it can be quantified and analyzed.

Let’s take the most straightforward example. A guy finally gets to spend the weekend with his girlfriend. So what should he do? This is a typical decision-making problem.

2. Three major factors in decision making

Element 1: Decision objectives.

Improve girlfriend satisfaction.

Factor 2: Decision-making level.

The highest level decision: whether to go out and play (yes/no)

Second level decision: Where do you want to go? (Decided: Go, then consider: Suburbs/Extreme Suburbs, Indoors/Outdoors)

Second-level decision: Which specific place do you want to go? (Already decided: far suburbs, outdoors, then consider: park/playground/attraction...)

Second level decision: How to get there? Where to eat? How to get back? (Has decided to go to a famous attraction in the suburbs, decided on the itinerary details)

Element three: Evaluation factors.

For example, based on the following factors, decide to go out and find a place to play:

1. It is autumn now and the weather is good.

2. My girlfriend likes to go out and play

3. There are several scenic spots nearby with good reputation.

4. I haven't been to these nearby attractions yet.

These are the evaluation factors that support "going out to play". Evaluation factors are an important basis for judging whether the decision is scientific.

Because it is very likely that the factors that decision makers consider during the planning stage do not hold true in reality, such as:

1. It is autumn now and the weather is good - these two days it has suddenly become gloomy and it is going to rain

2. My girlfriend likes to go out and play, but she is not in the mood these two days.

3. There are several scenic spots nearby that have a good reputation - but my girlfriend heard from her bestie that they are not fun

Therefore, it becomes very important to evaluate and revise decisions based on actual conditions.

3. Four principles of decision-making

There are a few basic principles to follow when making decisions:

Principle 1: Clear decision-making objectives.

If the goal is “I have to dump her as soon as possible”, then just think about how to use cold violence.

Principle 2: The decision-making hierarchy is built around decision goals.

If your goal is to "please your girlfriend", then don't think about staying home all weekend to play games and then using a Coke pull tab as a ring to fool people.

Principle 3: Decision-making at each level is constrained by the level above.

If you have chosen to go to the suburbs, your only options for transportation are: renting a car/taking a taxi, and it is best not to consider public transportation.

Principle 4: Each level of decision-making has its own evaluation factors.

Decision makers modify the evaluation factors based on actual conditions.

Now that we understand these four principles, we can look further into the role of data in decision-making.

4. How Data Supports Decision Making

Note: Decision-making is a business capability, and theoretically has nothing to do with data! Just like a guy going on a date, if he is tall, handsome, has an extraordinary temperament, and is very wealthy, then even if he does nothing, there will be a lot of girls coming to him, and the satisfaction is still very high.

BUT! Guys, look in the mirror and reflect on yourself three times a day: Am I tall? Am I rich? Am I handsome? Most guys don’t have stunning looks or a billionaire fortune, so they have to think seriously about how to get along with girls. At this point, more problems will arise.

Problem 1: They don’t know what to do. Many straight men don’t know what to do except to say hello. They don’t know how many restaurants, parks, amusement parks there are, or what movies there are. How can they make further decisions?

Problem 2: I know there are things I can do, but I don’t know how to evaluate them. I know there is Disney in Shanghai and Chimelong in Guangzhou, so I just drag my girlfriend there without considering the weather or whether I’m in the mood. As a result, I get criticized for nothing.

Problem 3: The evaluation level is not detailed enough. The more you look into the details, the more problems you have. My girlfriend wanted to go to Chimelong, so she excitedly booked a ticket, but she didn’t plan how to get there, what to eat, or how much money to prepare. As a result, she was hungry and tired along the way, and it was a waste of money.

When you encounter situations where you don’t know, are unclear, or have calculated the numbers incorrectly, that’s when data comes into play!

At this time, the data can:

• Clarifying the status quo
• Sorting out the logic
• Calculation process

Thereby supporting decision making.

The scientificity and accuracy of decision-making can also be guaranteed to a certain extent through data.

The so-called unscientific decision-making: You could have gone out to play, but you ended up staying at home and making your girlfriend unhappy. Or you could have gone to a tourist attraction, but you didn't go because you didn't know about it. These problems can be avoided by clarifying the current situation and sorting out the logic through data.

The so-called inaccurate decision: I should have taken a taxi, but I miscalculated the time and distance and ended up taking the bus, which made my girlfriend so tired that she got angry...

Although data cannot directly tell you what a scientific and accurate decision is, it is possible to evaluate whether the current decision is unscientific and inaccurate, and how likely it is to be scientific and accurate.

The story ends here. I guess many guys sighed: Oh my god, I don’t even have a girlfriend. This shows how difficult it is to make decisions! In fact, making decisions in a company is much easier than chasing a girlfriend.

5. Business Decision-making and Data Analysis

The reason why it is easier to make decisions in a company than to chase a girlfriend is that any company of a certain size has an organizational structure and personnel division of labor, and any formal company has a clear business scope. Therefore, as long as you understand the business characteristics of the company and the division of labor of the departments, you can figure out the decision-making characteristics by following the map. This is much easier than guessing what a girl is thinking.

From a data perspective, the core of corporate decision-making is not complicated. The core is:

• ABC indicators, which one should I do?
• From how much, to how much?
• How long will it take to complete?

Once these three questions are answered, the remaining issues are specific implementation issues. Leaders from different departments and levels will definitely care about different indicators.

Differences between departments:

• Strategic Development Department: Company-wide operational indicators
• Sales Department: sales amount, sales quantity, sales proceeds...
• Operations Department: total number of users, number of new users, number of active users...
• Supply chain: number of production orders, number of product supplies, number of materials used...

There are also differences in leadership levels:

• Company-level executives: strategic issues, such as development path, assessment criteria, and division of responsibilities
• Department-level leaders: tactical issues, such as which product lines to focus on, which customer groups to focus on, and task allocation
• Team and leader: Combat issues, such as whether to work in the morning or at night, colors red and blue, buttons on the left and right

Moreover, many decisions follow fixed routines. For example, for a single product, there are three typical strategies (as shown below).

For multiple product lines/business lines, there are also three typical strategies (as shown below).

Therefore, after accumulating data forms for various businesses, it will be easy to follow suit and help leaders clarify the decision-making logic (as shown in the figure below).

6. Difficulties in data-supported decision-making

In actual work, it is difficult to support decision-making with data mainly because:

1. Not understanding the business, and having no idea about departments, levels, division of labor, and goals

2. Failure to clearly distinguish the decision-making levels, either too superficial or too detailed.

3. The evaluation factors at each level are not listed, and the evaluation is not sufficient and cannot convince people.

Of course, the worse approach is to try to build a super awesome model of an invincible general, which can solve all kinds of problems. If there is such a powerful thing, I strongly suggest that you stop working and go directly to stock trading. You can be the next richest man in the world with a single touch. Yay! Sorting out the logic of the problem, advancing layer by layer, and combining multiple methods are the right way to support decision-making.

However, there are always those kind of girlfriends who are very pretentious:

• You asked her where she wanted to go, she said it didn't matter
• You suggested going indoors, she said it was boring
• You suggested going to the suburbs, but she said it was too tiring.
• You said let's discuss it, she said you are straight! No brain!
• Why can't you understand girls' minds?

I just don't say anything, but you guess my thoughts accurately. If you really meet a girl like this in a relationship, I strongly suggest you just kick her out to save trouble. We can't afford to serve her.

BUT!

If there really is such a boss at work, who treats algorithm engineers as fortune-telling engineers and expects "he won't say anything, and you can guess the things that determine the life and death of the company and he is 100% unaware of", then, besides quitting, is there a more elegant way to deal with it?

Author: Down-to-earth Teacher Chen WeChat public account: Down-to-earth Teacher Chen