Method and process of building a data indicator system [Template + case]

It turns out that the construction method and process of the data indicator system are inseparable from the " template + case " thinking mode. This article will show you how to build a good thinking system. At the same time, this article is recommended for those who are engaged in data operation analysis.

In order to improve students' competitiveness in the workplace, Teacher Chen has launched a series of articles, systematically explaining everything from the basic methods of data analysis to specific problem solving.

The students are confused in reality: I have seen the report you mentioned, and I update it every day. But how can this thing be a system? What if I have made a system? Why don’t I think what I have made is a system? Today I will give a systematic answer. To explain, let’s start with data indicators.

1. Why do we need data indicators?

Do you often hear the following:

• “There are probably more than 10,000 people.”
• “There are a lot of dissatisfied customers”
• “It feels like there is no one in our store.”

Uncertain, not specific, not accurate.

We usually talk like this in our daily life. There is nothing wrong with it, because specific information has a great cost, and most of the time we just talk casually. But if the business depends on this, then it will die. It is not clear how much money is spent and how much money is earned, and the boss will be so angry that he will die.

Data indicators are designed to combat this uncertainty.

If we change the above statement to:

• 9,800 new registered users on April 1, exceeding the target by 1,000
• On April 1, 100 items of product A were returned, and the cumulative return rate in 30 days was 2.5%.
• On April 1, there were 300,000 customers visiting stores nationwide, with a visit rate of 30%, lower than the expected value of 32%.

Isn’t it much more refreshing? This is the intuitive use of data indicators.

2. Why do we need a data indicator system?

In actual work, it is quite troublesome to accurately explain one thing. For example, if we want to say: "Product A sold very well in February!" If the other party wants to be serious, they can pick out a lot of faults (as shown below)

A problem often has many aspects, and only one indicator cannot fully explain the problem. This requires a set of logical data indicators to describe it, which is the data indicator system.

3. Five major components of the data indicator system

1. Main indicators (primary indicators)

The most core indicator used to evaluate how well something is doing. For example, "the product is selling well". The most intuitive indicator that comes to mind is the "sales amount", because this is the money we directly receive from selling the goods, so more money is better.

Each indicator must have the following elements:

• Business meaning: Its business meaning is...
• Data source: Which system collects the original data?
• Statistical time: The data generated within XX time
• Calculation formula: If there is a proportion or ratio, it is necessary to clearly state who divides whom; if it is a summary, it is necessary to clearly state who adds whom.

Note: You may need multiple main indicators to make a comprehensive evaluation. For example, if a product sells well, it is not enough to just look at the amount. You may also need to pay attention to the gross profit, which is the real money earned. You may also need to look at the sales volume, because the sales volume is directly linked to inventory, and you have to prevent too much backlog. In this way, there are at least three main indicators: sales amount, sales volume, and sales gross profit.

2. Sub-indicators (secondary/third-level indicators)

The main indicator may consist of several sub-components. For example:

Sales amount = number of users * payment rate * average order value

If the sales amount does not meet the target, we will be very curious: is it because there are fewer customers purchasing, or not enough sellers, or is it sold too cheaply? Understanding the details will help us find the real problem, and this is when we need to break down the sub-indicators.

3. Process indicators

The main indicator is often the final result. For example, the sales amount in the B2B industry is the final result of the series of processes: sales leads - pre-sales follow-up - demand confirmation - product experience - price negotiation - bidding - contract signing. It is impossible to supervise and improve the process by just looking at the final result. If you want to manage further, you have to look at it more carefully and add sub-indicators (as shown below)

4. Classification Dimension

It is possible that one thing is completed by many people over a long period of time. If you want to know how the total sales amount is composed and how much each region and each team has completed, you can add a classification dimension.

By using the classification dimension, the main indicator can be divided into several pieces, which can avoid the average trap and see the whole and the details together (as shown below)

5. Judgment criteria

Even with the above four points, we still cannot say: Product A sells well. Because good is an adjective, which is relative to bad. Therefore, we need a reference for comparison. The selection of a reference is a complex analysis process in itself, which requires in-depth analysis.

When constructing an indicator system, these judgment criteria are often presented together with the current data. This way, when looking at the data, you can make judgments intuitively, which is very convenient to use.

IV. How does the data indicator system work?

To summarize, the five parts of the indicator system are:

1. Main indicators
2. Sub-indicators
3. Process indicators
4. Classification Dimension
5. Judgment criteria

With these five parts, diagnosing the problem is very easy.

First look at the main indicator + judgment criteria . For example, if the main indicator is sales amount, first check whether the target has been met this month, and if it has not been met, how far it is from the target. Then check whether the annual cumulative target has been met, and how much deficit/surplus there is. This way, it is easy to see clearly: what the problem is and how big it is.

Let's look at the classification dimension . Which areas have not been done well, have they been doing poorly all along? Which areas have been doing well, have they barely completed or are they continuing to rise. In this way, it is clear at a glance who is capable of covering the bottom and who is lagging behind.

Then look at the sub-indicators/process indicators . Which link is not done well? Is it too few leads, so we need to increase promotion efforts; or the follow-up success rate is low, so we need to improve sales capabilities; or the quotation is always missed, so we need to increase some discounts. How to deal with the problem is clear at a glance.

It can be said that if a data indicator system is well established, 60% of the work of a data analyst can be done. A good data indicator system can allow business personnel to know at a glance where to work and in what direction to work, which is very useful.

Note: Diagnosis based on the indicator system only solves tactical problems, not detailed problems at the combat level.

For example, these questions:

• I have no idea, how should I design a method?
• I have three approaches, which one is more suitable for the current problem?
• I would like to use method 1, what are the chances of success?

These problems are easier to solve with thematic analysis. After all, financial statements are just tables that report the status of the situation. As for what to do in the future, more targeted analysis is needed.

5. How to construct a data indicator system

1. Clarify work objectives and main indicators

This is the most important first step. First, understand: what is the purpose of setting these indicators? Make the main indicator tree clear, so that you can know who to focus on when judging the criteria, and which sub-indicators correspond to which processes. As long as you work in a company, each department has its own KPI, so the main indicator can be found for sure.

2. Clear judgment criteria

This step is also very important, and it involves whether this is "a useful report" or "a bunch of colorful numbers." What is considered "good" is a very critical issue.

Now that the main indicator has been found, it is necessary to establish a matching judgment standard for it. This is the only way to interpret the meaning of the data and know how to look at the classification dimension. There are four common standards (as shown below).

Of course, setting standards is a complex analysis in itself, and it can be done in a very complex way. But in the end, it is necessary to clearly distinguish which ones are good and which ones are not.

3. Understand business management methods and find appropriate sub-indicators

With the main indicators and the criteria for judging the main indicators, we can further sort out the sub-indicators. The sub-indicators are directly related to the business management method.

For example, the sales amount can be broken down by branch or user. The specific way to look at it depends on how the business can manage it. For example, sales are generally managed by region, so it can be broken down by branch. The market is generally managed by user, so it can be broken down by user. In short, business convenience is the most important.

4. Sort out business processes and set process indicators

In theory, the more process indicators there are, the better. The more process indicators there are, the more detailed the process can be tracked and problems can be found. However, in business, data collection may not be done for every action, so it is necessary to combine the specific business process and control it at key nodes.

5. Add classification dimensions

There are many dimensions that can be used as classification dimensions. Which ones to choose depends entirely on the business perspective from which the problem can be managed. Add the dimensions that are meaningful to management. (As shown below)

In this way, a data indicator system is set up. The process is not complicated at all. Most of the time, the actual problem is: no data collection has been done, and there is no data record to make indicators. This is the most troublesome. However, the question is: why does the seemingly simple process not feel systematic when it is done?

6. Why I am not working on an indicator system

1. No main indicator, no idea what to do

This is the most common question. Many students receive reports from colleagues who have resigned. Why do they do it? Who do they show it to? What will happen if they see it? They don't know anything. Anyway, they just do it every day and update it regularly.

Some students try to figure it out, but the business side is confused. You ask them: What is your goal? They answer: To increase GMV. Dear, GMV is such a macro thing, what exactly does he manage? From what to what? How much is considered satisfactory? He makes his own plan by copying others, not to mention explaining it clearly to the data analyst.

2. There is no standard for judgment, and I don’t know what I said.

This is another common and fatal problem. Many students update reports blindly, listing a lot of data, but don't know what is "good" and what is "bad". Or they just naively think that: rising is good, falling is bad. As a result, many jokes are caused (as shown below)

3. Not breaking down the sub-indicators, staring at the main indicators

This problem is often a sequelae of departmental division of labor.

4. Not constructing sub-indicators according to business processes, simply piling up data

Many students like to pile up data when building a data indicator system. They put a bunch of indicators to make it look rich. But in fact, if you don’t find sub-indicators according to the business process, the logic between indicators is very poor, and it often looks inexplicable. Not to mention, it is easy to come up with weird things like "Are you happy?"

5. Not selecting the classification dimension according to the business and disassembling it randomly

Throwing in dimensions such as user gender, age, region, VIP level, source channel, terminal model, etc., the report appears to be very rich, but its actual business significance is unclear.

You asked him why he classified men and women, and he answered: There is a big difference after classification... As for what can be done if the difference is too big, and whether he has the business capability to do things according to gender, I don't know.

The above problems are essentially not considered from a business-useful perspective. They are simply set indicators for the sake of setting indicators.

This is directly related to work habits. Many students do not try to understand the business process and business goals by themselves, but instead look for an "authoritative", "formal", "perfect" and "universal" indicator system. The result is that they can only copy and paste from everywhere, and although it seems that they have done a lot, in the end, few people even look at the data.

If you want to change, you must start from the basics. Don’t think that just because you have a “data XX” title, you have to read all kinds of advanced algorithm theories to be useful. Theoretical work is for scientists, and when you work in a company, you have to do something practical and useful. If you want to help the business, you must start with serious research on the front-line business.

Author: Down-to-earth Teacher Chen

Source public account: Down-to-earth Teacher Chen (ID: 773891)