This is the best data analysis I have ever seen [Annual Work Plan]

How to do data analysis specifically? How to do it well? Let's take a look at the author's method~

It is the end of another year, and companies are making work plans for 2024. When it comes to data analysis work plans, many students are scratching their heads. How should a data analysis work plan be written? Today we will explain it systematically.

1. Common pitfalls of work plan

If other departments are asked to write work plans, the most likely style is as follows:

• Sales: Generate 2 billion in revenue for the company throughout the year, with a monthly target of XXX
• Operation: 10 events were organized throughout the year, and the sales volume on Double Eleven reached 500 million!
• Supply: Ensure the supply of goods with a revenue of 2 billion, and reduce the loss rate to 0.01%
• Development: Ensure the launch of 10 activities and the stable operation of the system for more than 300 days

So how to write data analysis?

• Method 1: Write 2,000 SQL lines every day and complete 500,000 lines in 250 working days per year
• Method 2: Build 20 prediction models and achieve a prediction accuracy of 99.99999%
• Writing method 3: Establish 10 data systems to promote the company's digital transformation
• Writing method 4: Provide 10 activity reports with 100% data accuracy

Question: Which of the above four ways of writing is OK?

This is related to the performance in 2024, so please stop and think for a minute.

answer:

What is written about sales and operations is related to the company’s performance and financing progress, and is directly related to everyone’s wallet!

Although supply and development do not make money directly, without them not a penny can be earned, so it is a rigid support.

The only thing data analysis does is neither rigid nor profitable, so it is dispensable.

Among the above four ways of writing, 1, 2, and 3 are seriously unsatisfactory. This is because 1, 2, and 3 are all about data analysis, which has nothing to do with performance and revenue. Although they say: data-driven, digital intelligence, and digital transformation, but who is driven, how much, and how to measure whether it is driven, nothing is clearly stated, and it is unknown whether the business department will admit it, it is just empty talk.

Only method 4 is barely acceptable:

1. At least put yourself in the position of a supporting department and have a clear positioning.
2. At least tie your work to the company's major projects so that it is not dispensable.
3. At least the results of the work are quantifiable (output 10 times), and when major projects are launched, it is necessary to look at the data.

Although it is still difficult to measure performance in this way, at least it brings yourself and the developers to the same level.

This is the starting point for the data analysis work plan.

2. Basic writing methods of work plan

Three iron laws of data analysis work plan:

1. Bundle work from other departments of the company.
2. Output content, New > Optimize > Ensure.
3. Quantify it in a way that other departments can feel.

Example:

After such optimization, the value of data can be reflected to a large extent, which is more useful than just writing: I did XXX. You should know that most people in other departments (including the bosses of most departments) don’t understand the principles of data. Having data or not having data, helping you make money or save costs, and being stable and error-free are the work results that most people understand better.

When making a plan, you have locked in the task goal, which makes it easier to do performance evaluation later. This fundamentally avoids the question of "What's the point of what you did!!!"

Of course, this is just the basic way of writing. If we delve deeper into the nature of data analysis, there are better ways to do it.

3. Advanced approach to work planning

Essentially, there are four steps from the generation to the use of data.

These four steps correspond to three important tasks of data analysis (as shown in the small picture)

The following parts are the key points:

Infrastructure construction will never be reflected as merit in any way.

If you do a good job, it's your job. If you don't do a good job, you're fired. This is the true status of infrastructure. So, if you take on infrastructure work, such as tracking points, designing/maintaining large tables used by business departments, setting indicators, and checking calibers, please be sure to tie it together with the company's key projects, major policies, multi-department linkages, and other major events! This way, there will be less resistance during execution, and it will be easier to calculate the credit when the credit is calculated.

For example, to improve the quality of the buried points, you should write:

• Sub-projects of the company's key growth projects in 2024
• Fill the gap of insufficient data in the original WeChat fission channel
• The total number of monitoring channels increased from 15 to 20
• Total user tags increased from 100 to 120

This way it is easier to quantify, so that everyone can feel the workload and it is easier to talk about it during the evaluation.

If you are not familiar with this writing style, just watch how the railway opening was reported in the news broadcast a few days ago:

• National 13th Five-Year Plan key projects...
• The total mileage is 1,500 kilometers
• The original 5-hour driving time was shortened to 2 hours
• Filled the gap of no high-speed rail from XX area to XX area

That’s the taste!

The core of data production is the word "tooling".

Without tooling, we don’t know whether people have read it or how much they have read it, and we can’t quantify the output. Among all the tools, new construction is always easier to show achievements than optimization, so we should list the goals of new construction first.

Work output, the more conspicuous the better:

• The data screen is more eye-catching than the data dashboard
• Data dashboards are more prominent than data reports
• Data reports are more prominent than Excel reports
• Excel is more conspicuous than sending a few numbers via email

Therefore, when making work plans, communicate more with the business department to collect information about major events and projects next year. Then, try to promote prominent data products, and push other complex needs when necessary, and recruit people when necessary. This will make it easier to show achievements.

The core of data usage is the scenario.

The more specific the scenario, the more likely people will use it. If you make a report for sales, it is estimated that less than 1% of people will read it. If you directly send task reminders on WeChat for Enterprise, even those who don’t click on them will click on them. The reading rate will go up sharply, and it will be easier to evaluate the results.

If you want to make a forecasting model for the supply chain with 100% accuracy, even the gods can’t do it. If the scenario is specific to: reduce the cost of wrong selection, it is estimated that there is a chance to pick out the obvious failures;

If you have to exhaust all the user characteristics when creating user portraits for operations, it will probably be useless. However, if you are specific about screening high-potential users and eliminating the wool party, you may only need a few characteristics to solve the problem;

So if you want the data to be useful, you have to discuss the scenarios in great detail. The more detailed the scenarios, the better. It is best to have four or five application scenarios for a set of data, so that you can maximize the benefits. If you make a detailed plan, you will have a lot of things to write about when evaluating performance.

4. The core difficulty of data analysis and planning

What is the core reason why data analysis planning and assessment are difficult?

Everyone said verbally:

• Digital transformation is really important
• Data analysis is very useful
• Data-driven business

But when it comes to performance evaluation, we ask:

• The company made 500 million yuan on Double Eleven. How much of it was made by your model and how much was made by others?
• You can write SQL, so can developers and operations guys. What’s your special contribution?
• How much more money can the salesperson earn by looking at your report or not?
• There are so many things to be digitized, can you digitize them just by giving a number?

This is the root of all the difficulties in data analysis! Therefore, when making a plan, you must carefully sort out the scenarios and choose the right direction, so that everything goes smoothly during the assessment. Otherwise, if the direction is wrong, the plan will become a running account. Naturally, there will be no good results.

I guess after reading this, students will feel that the difficulty lies in the use of data. How can we lock in the scenarios so that the business departments can truly use it and acknowledge that this is the result of data analysis?

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