How should operations use data?

How should operations use data?

This question stems from the following two questions:

• What is data operation , what does data operation do, and what kind of support does it provide?
• What operations provide to BI is too diverse and messy, and often meaningless after completion. Are there any techniques to make the work more effective?

Data is a basic skill

Let’s not talk about data operation for now, data is the basic skill for any Internet job. Basic skills mean that operations deal with data every day, KPIs are expressed in data, operations strategies are adjusted based on data, and all operations work is driven by data.

Therefore, looking at data, analyzing data , and proposing ideas based on data are things that every operator must do.

So, what is data operation?

Data operations means that everything is carried out around data.

First of all, data operations need to understand what data is needed in the process of product operation . For example, in e-commerce , we must first look at the order volume, average order value, and conversion rate , and also the process data of users flowing through different pages, where they stay, where they scroll down, and so on.

Secondly, data operations must define the meaning of data. For example, "activation" in an app is defined as whether the user downloads the app and completes registration, or whether the user uses a certain function.

Then, it is time to create a data viewing path. Is it a daily email report or visualization?

Then, in different companies, there may be different changes in responsibilities. Some companies' data operations are biased towards BI, which is to extract data, analyze data and form reports, and even do modeling, etc. Some are biased towards operations, which is to discover problems and opportunities from the data, and propose and verify hypotheses. Some companies are biased towards the market, which is to do industry analysis and combine it with the company's business development level.

Data operations are usually separate because:

1. The existing team members have bottlenecks in data mining capabilities;
2. The business type is rather special and requires an independent team.

Only in these two cases will an independent data operations team emerge.

How to make data request

As the person who probably deals with the data department the most, operations should learn to correctly propose data requirements.

First of all, if there is a data demand process within the company, then follow the data process first. When proposing data requirements, you need to make a few things clear:

1. The time dimension of the data. What is required is the data within a certain period of time, this needs to be made clear.
2. The type of data to exclude. It should also be made clear whether the required data includes items that need to be excluded.
3. Detailed presentation of data. What data is required and what dimensions it contains (note, not necessarily fields) should be listed.

For example, I am an e-commerce operator and I want to understand the behavior of newly registered users in the past week. So maybe when I make a request it might be like this:

Data pull objects: users who registered between November 1, 2017 and November 7, 2017, with the accuracy required to be between November 1, 00:00 and November 8, 00:00

Analysis content: A wide table is required to reflect the browsing and ordering status of the selected objects in the App during this period. The fields are as follows:

Of course, this is just an example.

With this wide table, you can use the data pivot to know which product types are more popular and the behavioral preferences of newly registered users in that week.

Second, once you have the data, you should do two things:

1. Thanks to the person who helped you get the data
2. Use data to conduct analysis and complete data reports

When you complete step 2, you should do two things:

1. Give the report to whomever you wish;
2. Feedback the results of data analysis to the person who provided the data.

The provider of data actually cares very much about whether the data he/she provides creates value. We must not let him/her down, otherwise the consequences will be serious.

Notes on Data Analysis

When using data as an operator, there are some things to pay attention to:

1. Don't ask for data with conclusions. To a large extent, you will misinterpret the data itself because of the misleading conclusions you have drawn.
2. Collect as much intelligence as possible about the impact of the data, but also have a certain imagination to explore the possible reasons behind the data.
3. Try to choose desensitized data when making data requests and using data.
4. It is workplace etiquette to require feedback to the data provider when receiving data.

That's all, hehehe.