Indicator dictionary and indicator system are both very important aspects of a data analyst's work. Today I will mainly share content related to the indicator dictionary.

01 What is the indicator dictionary

"Lao Liu, our turnover this month is only 1 million. We need to keep working hard!"

"What? We clearly sold 1.5 million this month, how come it's only 1 million? Is your data wrong?"

"That's impossible. I got it from the database myself. One million!"

"I checked the number, it's 1.5 million!"

02 The value of indicator dictionary

(1) Reduce communication costs and improve communication efficiency

(2) Break down information barriers and reduce duplication of corporate construction

(3) It is the foundation for the company’s digital construction and the foundation for building a data platform

03 What are the parts of the indicator dictionary?

A standard indicator dictionary actually consists of two parts: the indicator part and the dimension part . These two parts are dictionaries when separated; when combined, they can generate various indicators covering daily business use.

(1) Indicator section

Let me give you an example. The following are some of the indicators in the indicator dictionary in Baidu Statistics:

This is presented externally, so it simply includes two parts: indicator name and indicator definition. But as a complete indicator dictionary, it must also have the following parts.

• Indicator types: such as basic indicators (the most primitive simple indicators, which cannot be further subdivided, such as the number of orders and order amount), composite indicators (generated through various calculations based on basic indicators, such as order rate = number of orders placed/number of added purchases).
• Limitation: Describes the limitation of the indicator, such as limiting the number of users to new users
• Limited dimension: describes the dimension that the user must limit when querying the indicator, such as time.

(2) Dimension part

Dimension is the angle of analysis and the direction of splitting.

To make it easier to understand, let’s first give an example. Still Baidu Statistics.

These dimensions are actually some of the dimensions commonly used on the Internet.

When indicators are superimposed with dimensions, various indicators that meet business scenarios can be generated. For example, the most commonly used dimension is the time dimension, "transaction amount in the past 7 days". "Past 7 days" is the time dimension, and "transaction amount" is the indicator. It is also possible to overlay indicators in multiple dimensions at the same time. For example, “the number of iPhone orders placed in the past 30 days.” How to split it? It’s very clear, right?

Of course, not all dimensions + indicators are valuable. How to generate valuable dimensions + indicators is what the indicator system will share later.

How to build an effective indicator dictionary

The above describes the value of indicator dictionaries and the contents of indicator dictionaries. It doesn’t seem to be troublesome. Isn’t it enough to just sort out the indicators and dimensions and write a document? In fact, it is not that easy.

The difficulty lies in implementation. If you compile a dictionary but put it on the shelf, it's the same as not having it at all.

So how do we derive an effective indicator dictionary?

(1) It must be tailored to the business application scenario and start from the business rather than working in isolation

Applications that meet business needs are the most fundamental starting point. For example, if the company as a whole is concerned about transactions, then the focus of the indicator dictionary should also be transaction-related; if the business is concerned about service experience, then the focus of the indicator system should be service experience.

(2) Communicate fully with each business department and strive to reach a consensus within the company

Many times, the reason why it is difficult to promote a unified indicator dictionary is that different businesses want to use calculation indicators that are beneficial to them. This requires a balance. Some top-down push is also needed. It is difficult to convince the business to calculate performance and perform statistical analysis according to your specifications.

(3) Maintain the indicator dictionary and promote its application

Maintenance is also difficult. Because new indicators and new demands are constantly emerging, updates and maintenance must be carried out on the original basis. At the same time, training should be carried out so that the entire company can speak with a set of standards.

That’s all for now. Welcome to communicate.