I finally made it clear what the underlying logic of data-driven growth is

Introduction: Growth effects always make people feel difficult. This article describes the methodology of data analysis-driven growth. Each point is an in-depth summary of the author's past experience. I hope it will be helpful to data analysis partners.

"Data analysis should help growth!" is the requirement of many companies. However, when it comes to actual implementation, many students are stumped. It seems that the daily work is just to calculate data, how can there be growth? Some cases talk about AB testing, but the version is produced by the product, and the fission activity is done by the operation. I just calculated a data.

Today I will explain to you in detail how to achieve growth.

1. The underlying logic of growth

Let me ask you a soul-searching question: Why does the business grow? To grow, first of all, the business itself cannot be too bad, and secondly, the business must have enough room for development.

Under these two premises, four underlying logics of growth emerge:

1. There is no need to change the present situation. Simply increasing investment will lead to greater output.
2. We need to make changes to the present and fill in the weakest link.
3. We need to make changes to the present and promote the most powerful practices now.
4. Why not just stop doing what you are doing now and find a new field with more opportunities?

A picture can be summarized as follows:

So, the next question is: What can data analysis do in this process?

2. The underlying logic of data assistance

The second soul-searching question is: Can we really not do business without data? Of course not, we can still do business without data. You can see that this is how it has been done in history. In history, wars were fought with "countless enemy soldiers and countless enemies killed". In history, cooking was done with "a little, a proper amount, a little, a moment". It is completely possible to do business without data - we just don't know how well we do it.

Then comes the third soul-searching question:

• Do you prefer "countless enemy soldiers" or "500 enemy soldiers"?
• Do you prefer "a pinch of salt" or "5 grams of salt (a spoonful of salt from a salt shaker)" when cooking?
• Would you rather be asked to "Wait a moment" or "Just wait 15 minutes"?

Everyone prefers precise judgments, clear results, and visual processes, which are all where data comes into play.

Therefore, if you want to do a good job of data assistance, you must have a clear mindset from the beginning. Don’t expect data analysts to come up with a super-powerful event plan or a super-cool product prototype. Instead, use data to accurately judge, visualize the process, and test the results. It is best to output in the form of products. In this way, all combat troops can be equipped with maps, radars, and drones, thereby improving their overall combat capabilities.

3. Analysis of additional investment

Some students began to wonder: additional investment, this business will also cost money, give me 500,000 yuan and I will do 2 million, give me 1 million and I will do 4 million. Does this need to be analyzed?

This really requires analysis, because input itself involves several types (as shown in the figure below), and each type of input can increase output.

Sales is a domino effect, and if you invest in the wrong direction, you will easily run into bottlenecks. Common examples include:

1. Excessive sales and insufficient production lead to out-of-stock
2. Over-investment in sales, a surge in production, leading to negative reviews/returns/complaints
3. Over-investment in marketing results in no direct output and wastes money
4. Excessive investment in marketing, serious cross-subsidy, insufficient output, and serious waste
5. Over-investment in marketing results in bottlenecks in sales channels and no output
6. Over-investment in supply resulted in poor sales and serious product backlogs
7. Investing in sales, marketing, and supply together resulted in a huge cost explosion and a broken capital chain.

Therefore, even a simple decision of adding more investment requires comprehensive consideration of multiple factors:

1. Target market potential
2. Sales input-output ratio
3. Incremental effect of marketing activities
4. Supply chain expansion capabilities

Each calculation step here requires the support of several small analysis points. For example, the target market potential needs to consider the target user base, target user purchasing power, competitive product penetration, and the proportion of users covered by the brand. This requires the support of industry analysis. The incremental effect of marketing activities also needs to consider the output level under non-activity, the superposition effect of activities, the effect of activities on future demand, and the wool-pulling effect brought by activities (as shown in the figure below).

Therefore, a lot of analytical work is essential, and it is not a simple decision like "give me 50 and I will do 100".

4. Benchmark Analysis/Problem Analysis

Many students are familiar with the method of finding benchmarks/problems in existing businesses.

• If you only look at income indicators, use stratified analysis to separate high, medium and low
• If we combine the income and cost indicators, we can use matrix analysis to screen out individuals with both excellent performance.
• Then, use the funnel analysis method to find the worst performing link in the business process

The question is: after distinguishing between good and bad, what should we do?

• For benchmarks, the core issue to be addressed is: Can the benchmark really be replicated?
• If the benchmark is replicable, spread the experience;
• If benchmarking can only be successful in a specific market environment, then it depends on where this environment exists.

If the benchmark is that only certain people/products/channels can succeed, then go find people/products/channels of the same type.

To determine replicability, we must first conduct a life cycle analysis + feature analysis of the benchmark

From the perspective of the life cycle:

• Is there a special environment during the benchmark startup period?
• Is there any special support during the benchmark development period?
• Is the benchmark development stable or just a flash in the pan?

From the perspective of characteristics: Are only people in special locations/special types of goods/special physiological characteristics able to succeed? The collection of this intelligence is very important. It is necessary to review historical data, collect market information, and collect historical business actions. Only by labeling stores/goods/people can a full analysis be achieved.

Problem analysis: After finding the problem, the important thing to solve is whether the problem can be overcome. In the short term, it is very likely that many problems cannot be improved. Therefore, is there a plan to improve the problem? What specific points have been improved? What is the effect after the improvement? Detailed data records should be made so that in multiple rounds of comparison, the answer can be: whether the problem can be improved. If it can be improved, then break through the bottleneck. If not, consider bypassing it.

5. Analysis to discover new opportunities

What are the new opportunities?

1. Although a channel has less traffic, its conversion rate is higher than others
2. Although a certain type of customer is small in number, their ability to pay is stronger than others
3. Although a new region or new category has just started, its growth rate is faster than others.

These are all potential opportunities. It is not difficult to find them out. You can see them through stratified analysis + cohort analysis.

The difficulty of the problem is: what happens after finding it? Is it really possible to expand the new opportunity point? Are the existing means really effective? Will the benefits be diluted after it becomes bigger? We don’t know any of these. So more in-depth analysis is still needed.

Here, there are two approaches: one with a business strategy and one without a strategy. If a business strategy is already in place, then you can directly monitor the effect and observe whether it is getting bigger and bigger and whether there is diminishing marginal benefits. If there is no strategy, then you need to combine historical situations and make full opportunity insights. For example, to find opportunities in a certain user group, you need to look at:

• How many users in this group can be acquired through existing channels?
• Which channels have a high concentration of users in this group?
• Is the high-concentration channel special? Is there room for improvement?
• For existing activities, is the response rate of the group high enough?
• For existing products, which of the following groups has a higher first purchase rate or repeat purchase rate?

Only with sufficient potential base + highly responsive products/activities can the "opportunity point" be supported. If not, it can only be handed over to the business to consider whether to arrange a testing plan and start with the test results.

summary

From the above, we can see that if we really want to drive performance improvement, we need data analysis around "growth", do a lot of auxiliary work, combine industry data, historical data, current performance, test results, truly interpret the key to growth, accumulate experience/lessons, and then we can achieve it.

In many companies, data analysis is simply calculating channel ROI, crossing cities/stores with sales, and seeing which is higher and which is lower. This is definitely useless. For more data-driven business methods, please visit Knowledge Planet for in-depth learning.

Author: Down-to-earth Teacher Chen

Source: WeChat public account "Down-to-earth Teacher Chen"