Complete! [Business Analysis Model] Construction Guide

In data analysis and business optimization, building an effective business analysis model is the key to achieving data-driven decision-making. This article systematically introduces the construction method of business analysis models, and explains in detail the application scenarios, construction steps, and how to optimize each model based on actual business needs for your reference.

Many students are most afraid of hearing the word "build a model". Especially building a "business analysis model":

1) How to combine 4P, 4C, SWOT and the like with data?

2) LR, SVM, CNN, etc. are far from the business

When it comes to sales, products, operations, logistics and other business work, how should we come up with an analysis model? Let's explain it systematically today.

01 What is a business analysis model?

Business analysis model, that is, through fixed indicators + fixed analysis dimensions, to reflect business conditions and diagnose business problems.

The key to building a business analysis model is to understand:

1. Does the business understand the current situation?
2. Do business colleagues understand the source of the problem?
3. Is there a solution to the business problem?
4. Is there any prediction on the business development trend?

Obviously, different people have different states. So before modeling, you need to sort out the following:

1. What level is the department I want to serve?
2. What are their most pressing issues at the moment?
3. Is the business process clear? Has enough data been collected?

Students can make clear judgments as shown in the figure below.

Accordingly, there are four commonly used models:

1. Current situation description model
2. Question Classification Model
3. Business Optimization Model
4. Business forecasting model

Let me introduce them one by one below.

02 Current situation description model

Current situation description models are used in the early stages of business development to describe business performance using multiple indicators, thereby establishing a business monitoring system. The commonly used AARRR, PRAPA, RFM, funnel models, etc., all belong to this category.

The current situation description model is derived from the business process.

There are two common processes: series and parallel.

1. Serial: A process that is divided into n stages, and you need to complete one step before you can move on to the next step. It starts from the beginning of the process and ends at the end.
2. Parallel: refers to a task being completed independently by each line at the same time. Starting from the overall goal to the end of the task

Most business processes are a combination of these two modes (as shown below)

The biggest function of the current situation description model is to clarify responsibilities and expose problems.

When there is a problem with the main indicator (such as sales revenue, production cost), you can trace back along the business process to see which link has the problem. Therefore, it is often used in sales management and operation management.

But please note: current situation ≠ problem, current situation + standard = problem. Therefore, only when the standard is single and clear can we directly see the problem.

If the standard itself is complex, further measures are needed.

03 Question Classification Model

The problem classification model is used to determine whether a business problem occurs based on multiple indicators.

If there is only one criterion for judging whether an indicator is good or bad, such as cost or profit, then there is no need for a model. Just check whether the single indicator meets the standard.

However, when there are more than two indicators, a comprehensive evaluation method is needed, which is the "problem classification model". If two standards are needed to judge whether a business is good or bad, and the correlation between these two standards is low, a matrix model can be used for classification.

The common important emergency matrix, Boston matrix, and quality/quantity matrix all follow this principle (as shown below).

If the number of judgment criteria increases to more than three, and there are too many cross-judgment criteria, it will be difficult to judge who is good and who is bad by naked eye. At this time, DEA method or AHP can be used for judgment. Compared with pure machine learning method, DEA method is simpler and more direct, and AHP method involves experts, which are more easily accepted by the business.

04 Business Optimization Model

Business optimization model is an analytical model that finds the optimal configuration when input and output levels are limited by resources. For example, given the labor cost of each department, find an optimal task allocation.

The most common one is solving linear programming, which is very commonly used in work allocation (as shown in the figure below).

There are similar scenarios on the marketing side, such as:

1) How should each sales team allocate leads/sales expenses to achieve sales targets?

2) To achieve the highest customer acquisition efficiency, how should the costs be allocated for each delivery channel?

3) How should resources be allocated to each product line to maximize overall gross profit?

In short, any business combination optimization involving multiple businesses is suitable for this type of model.

05 Future prediction model

Business forecasting models predict business trends based on past data. The basic assumption of all forecasts is that the laws of the future will be the same as those of the past, and the past scenes will reappear in the future.

Therefore, when making business forecasts, they often assume that some business parameters are fixed and then speculate on future situations.

Common practices, such as:

1. Assuming the life cycle trend remains unchanged, speculate on the future situation
2. Assuming the conversion rate/retention rate remains unchanged, speculate on the future situation
3. Assuming the input-output ratio remains unchanged, speculate on the future situation

In some stable industries, these assumptions are often accurate. But please note that there are three situations in which the assumptions may fail.

1. New businesses and new scenarios mean there is no historical data to refer to.
2. The sudden and unclear circumstances caused all conversion rates to be abnormal.
3. There are obvious problems with business operations, and it is no longer possible to reverse the normal trend.

At this point, it is recommended to conduct more tests to obtain reliable parameters rather than blindly predict. After the business model is established, there is no need to start from scratch every time you encounter a problem, sort out indicators, list hypotheses, verify ideas, find inspiration... You can apply it directly, which greatly improves efficiency.

Moreover, these four models are progressive.

As the business develops from 0 to 1, by establishing the current situation description model → problem judgment model → business optimization model → business prediction model one by one, we can achieve increasingly efficient data-driven development.