What does data operations do and how should analysis be done?

This article takes an in-depth look at the daily challenges faced by operations staff and the conflicts between them and data analysts, while also providing effective solutions and suggestions to help data analysis better support operations and drive business development.

The easiest way to do data analysis is to conflict with operations.

First, there are too many operational data requirements, and they are often urgent and bizarre;

Secondly, no one reads the reports given by data analysts. Operations officers prefer to run data and write reports by themselves, and a special position has been derived from this: data operations (although this was not the original intention of data operations, in many companies it has been turned into operations staff who write SQL).

How should data analysis be done to support operational iteration?

1. What does operations do?

Students who may work in operations will say with deep conviction: operations is a job of miscellaneous tasks.

However, despite all the complaints, operations is an important part of Internet companies, and is known as the iron triangle along with products and development.

1. Product: design product and observe product effects
2. Development: formulate product requirements and maintain product operation
3. Operations: Do everything except the above two things

for example:

• Attract users to use the product: promotion operation
• Give users a small gift: User operation
• Send a few messages to users: content operation
• Organize an event to attract users: event operation
• Prepare products for users to buy: Product Operation
• WeChat, Weibo, and Tik Tok: New Media Operations

…

After-sales processing can be handed over to operations in Internet companies. Operations do everything! This is why operations seem to be a chore. Small companies have few operations people, and only a few brothers go to heaven and earth, go to the sea and catch things, and do everything. Operations in large companies will have a slightly clearer division of labor and promotion path.

The output of operations is very important. As the saying goes: If the product is not good enough, the operations will make up for it. Especially in China, Internet products are basically copied from each other. If homogeneous products want to compete, they can only rely on differentiated operations.

for example:

• Let's discuss a hot topic: content/new media operations
• Send a coupon to attract you: activities/users/promotion operations
• Prepare for hot-selling products: Product operations

It can be said that the most popular things in life (saving 5 yuan today, receiving a gift tomorrow) are all done by operations. However, why does such a popular operation always conflict with data analysis? This starts with the relationship between operations and data.

2. Operational data analysis from the perspective of laymen

Operations inherently require data. To make up for product deficiencies and improve user conversion, operations must have clear data guidance:

What is the current user and product status?

So what if we're done?

These are the two core themes of operations.

In the eyes of laymen, operational data analysis is nothing more than AARRR and the funnel model. AARRR is an indicator system used to describe user status, and its basic structure is as follows:

The funnel model is often used to improve content delivery, new user registration, and product processes. These tasks are often a combination of several links. Users will be lost during the operation due to too many links, which is similar to the effect of a funnel, so there is a funnel model.

Operations also like to mention user portraits. Because writing copy, organizing activities, and sending gifts all involve the question of "what do users like?", so many operations hope to have a user portrait to help them see the characteristics and preferences of users.

These three requirements are clear when viewed individually. However, in actual use, operations always like to raise a lot of temporary data retrieval requirements, and each version is different.

Why? Because students who understand data can see the problem at a glance: AARRR, conversion funnel, and user portraits are essentially descriptive statistics that tell you a result. These data requirements can be fixed as long as reports are made.

But the real pain point of operation is not that I don’t know what the result will be, but: the result is like this, and my leader is not satisfied no matter how hard I try, what should I do!

3. The real pain points of operation

Pain point 1: No money

"Pinduoduo's Latest Strategy" - Did you give them such a big discount?

"Luckin Coffee's Top Ten Success Stories" - Can you give them so many coupons?

"Didi's Golden Rule for User Growth" - Can you give them so many subsidies?

If you don’t have money, you can’t say anything. However, there is a group of leaders who like to read “How I got 10 million fans at 0 cost”, “How to fission 5 million fans at 0 cost”, “Growth can be achieved with 0 investment”. In short, there must be a way to do it without paying money. If there is no way, it’s your problem. You want to…

Pain point 2: We have money but no effect

For example, new users rarely spend money when they register, so we decided to do a "10 yuan discount on first order after new registration". Since 10 yuan is already off, it's time to buy it. The result (as shown below) is:

Pain point 3: It works, but leaders are demanding more and more

For example, new users rarely spend money when they register, so we decided to do a "10 yuan discount on the first order after new users register" activity. The effect was very good, and the first order rate of new users increased rapidly after registration.

As a result, the leader was not satisfied again:

“I’ve got the first order, but the second purchase doesn’t seem to be possible.”

“There is secondary consumption, but it is all small orders, so GMV cannot be increased.”

"Since we have invested so many resources, we must achieve our goal in one fell swoop!"

"Think about it more, grasp the underlying logic and core thinking, don't just throw money at it"

…

Everyone who has done activities/user operations will burst into tears when seeing this. In short, as consumers, we see those activities that are like the Olympic Mathematics, "first deposit, then inflate, then cap, then combine, then fission, then withdraw", where the rules are not clear at all, are all forced out like this...

Pain point 4: It works, but the leader says “You can’t just walk on one leg”

Finally, the activity was effective! However, the leader was not satisfied again:

You see, newcomers all rely on these activities. What will they do if there are no activities?

Look, PDD has a new way of playing, we can't just stick to the old ways

Have you read the article "Shocking! Alibaba's latest technology revealed"? Why don't we make one?

In short, only one form is not enough, keep innovating. Innovation requires looking at data, doing calculations, analyzing needs, designing rules, testing results, and if the results are not good, keep struggling: how to innovate.

Pain point 5: It works, but now I have no money...

The activity was very effective and the number of participants kept increasing. However, now there are too many new members and the expenses cannot be sustained (as shown in the figure below). So the leader suggested: come up with some innovative ideas and don’t always use the same old routines, as the costs cannot be sustained.

Tangled, tangled. This endless tangle is the root of all evil.

However, the essence of operation is so entangled:

If you want to raise a certain indicator in a short period of time, you can do it by throwing money at it. But you may not always have money to invest, so you will be entangled.

Spending money always brings about related problems and always causes sequelae, so there is always a need to balance all aspects and to struggle with them.

Even if it works temporarily, in the long run, user interests will change and costs will also have a ceiling, so innovation is always needed and the struggle is endless.

Innovation may fail, and we have to start all over again.

IV. Summary

Obviously, simply stating the current situation and simply presenting a bunch of data will not help solve the above problems. Even when the general direction is not determined, the more detailed data is presented, the more it will lead people astray.

If data analysis really wants to help operations get out of trouble, it must first be able to interpret what operations are worried about. Start from real scenarios and think about countermeasures.