When we want to innovate, we often need to have practical evidence to support our ideas. Business consultants are often seen as smart people with a modeled, analytical mindset that helps them better understand markets, competitors, and customer needs. Business analytical thinking is a systematic way of thinking, which can help us better understand business problems and come up with practical solutions.
The following are 8 common business analysis thinking, which may not instantly upgrade your thinking mode, but may bring you a "flash of inspiration" feeling for your future work.
01. Classification model
In our daily work, we need to carry out classification thinking , such as customer grouping, product classification, market classification, etc. The classified results need to be able to distance in the core key indicators, that is, the classified objects must have a significant tendency to cluster, rather than random distribution. Usually, we will use the core indicators we focus on as the horizontal axis and the vertical axis, and classify based on these indicators.
For example, the RFM model is a commonly used user grouping system, which is constructed based on three core indicators of charging (last consumption time, consumption frequency, and consumption amount). We divide the actual users into eight regions according to these indicators, and each region represents a different user value group. We need to promote the transfer of different users to more valuable areas, and adopt different strategies to increase their value according to the value of different paying user groups.
On the three indicators of R/M/F, we have divided the actual users into the following 8 areas through experience (as shown in the figure above), and what we need to do is to promote the transfer of different users to more valuable areas. That is, each paying user is matched to different user value groups based on consumption behavior data, and then different strategies are adopted according to the value of different paying user groups (as shown in the figure below)
Classification is a common method that can be applied to many scenarios, such as user classification (such as new and old users, activity, consumption power level, etc.) and product classification (such as price band, specifications, user needs, etc.). In data mining or machine learning, classification problems occupy a large proportion, and its purpose is to group objects with certain commonalities or similar characteristics into a group for better management and fine-grained business operations. In addition, classification also helps to study the commonalities and differences of similar things in order to better understand their characteristics and user needs. But it should be noted that classification is only a means, and we should not classify for the sake of classification.
02. Quadrant model
Quadrant models are an evolution of classification models that are no longer limited to classification using quantitative metrics. When we have no data support, we can only infer subjectively through experience. In this case, we can combine some important factors into a matrix, roughly define the direction of good and bad, and then analyze it. The famous Boston matrix is a classic management analysis quadrant model, which can help us classify and analyze.
Quadrant analysis has the following advantages:
-
Find common causes of problems: Through attribution analysis of events with the same characteristics, common causes can be summarized. For example, in the case of advertising promotion, the events in the first quadrant can extract effective promotion channels and strategies, while the third and fourth quadrants can exclude some ineffective promotion channels.
-
Establish group optimization strategies: For different quadrants, you can establish optimization strategies. For example, in the RFM customer management model, customers can be divided into different types such as key development customers, key maintenance customers, general development customers, and general maintenance customers according to quadrants. Then, more resources can be given to key development customers, such as VIP services, personalized services, additional sales, etc. Sell potential customers a higher value product, or offer some incentive to entice them to come back.
03. Funnel model
The funnel model is a relatively well-known analysis method. These include long funnels and short funnels. Long funnels usually involve more links and a longer time period, such as channel attribution model, AARRR model and user life cycle model. Short funnels have a clear purpose and are short in time, such as order conversion funnels and registration funnels.
Funnel analysis is usually used to analyze the link analysis of multi-service links and user transactions. However, the length of the funnel needs to be paid attention to when applying it. The funnel should not exceed 5 links, the percentage value of each link should not exceed 100 times, and the conversion rate value of the last link should not be lower than 1%. If these limits are exceeded, it is recommended to divide the funnel into sections for observation.
This is because if there are more than five links in the funnel, there will often be multiple key links, and it is easy to analyze multiple important issues in one funnel model, resulting in confusion. In addition, when the magnitude difference between the percentage values of each link is too large, it is difficult to detect the fluctuation relationship between the values, and it is easy to miss information. Therefore, attention to these limitations is required to avoid confusion and omission of information.
04. Pareto model
The Pareto model (also known as the ABC analysis model) is derived from the classic 80/20 rule, that is, 20% of the data produces 80% of the effect. In data analysis, using the 80/20 rule can help us find the focus and improve the effect.
In business operations, we often need to prioritize. The ABC analysis model is a tool that helps us make this decision. This model can be used to divide products, customers and customer transactions. For example, in terms of product classification, we can arrange the sales in descending order and calculate the percentage of each product's sales in the total sales. We can then divide the products into A, B, and C classes based on the percentage difference. Products with percentages within 70% (inclusive) are classified as Class A, products with 70% to 90% (inclusive) are classified as Class B, and products with 90% to 100% (inclusive) are classified as Class C. This division can help us understand which products contribute the most to sales and thus better allocate resources.
The ABC analysis model can be used not only to classify products, but also to classify customers and their transaction volume. For example, we can divide customers by calculating each customer's contribution to the total profit. If only 20% of our customers contribute 80% of our profits, then we know we need to focus on maintaining these customers. This analytical model can help companies understand which customers or products contribute the most to profits, leading to better strategy and decision-making.
05. Logic tree analysis method
Logic tree analysis is a tool for decomposing a problem into multiple sub-problems, also known as problem tree, deduction tree, and decomposition tree. By detailing the problem layer by layer until a solution is found, logic tree analysis ensures nothing is missed or overlooked and builds consensus. This approach is widely used to solve business and social problems.
We can take a known problem as a tree trunk and start thinking about which related sub-problems or sub-tasks this problem is related to. Every time you think of something, add a branch to the problem (that is, the trunk) to indicate what problem this branch represents. In addition, the branch can be further forked, and a large branch can have multiple small branches, and so on to find out Questions for all related items.
This is the logic tree analysis method, also called problem tree, deduction tree or decomposition tree.
06. Retention/Queue Model
Cohort analysis is to slice the observed objects in time according to certain rules to form an observation sample, and then observe the changes in certain indicators of this sample over time. Currently, the most common application scenario for cohort analysis is retention analysis. Retention analysis refers to the analysis of user activity within a certain period of time to understand their loyalty to a product or service. Cohort analysis can help us understand the changing trend of user retention rate, and formulate corresponding strategies according to these trends to improve user retention rate.
Retention rate refers to the proportion of new users who continue to visit, log in, use or convert after a certain period of time. The retention rate can be divided into three categories according to different time periods. Taking login behavior as an example, the retention rate refers to the proportion of new users who still maintain login behavior after a certain period of time to the total number of new users at that time.
What the retention rate reflects is actually a conversion rate, that is, the process of converting initial unstable users into active users, stable users, and loyal users. With the change of statistics, operators can see the changes of users in different periods, so as to judge the attractiveness of products to customers. From the product point of view, through the previous retention analysis, we find the key behaviors that trigger the retention, help users find the key nodes of product retention as soon as possible, and improve the retention of early users.
07 Association Model (Market Basket Analysis)
I believe you should have heard such a case: In supermarkets, baby diapers and beer are often sold together. The reason is that after data analysis, it is found that fathers who buy diapers, if they buy diapers at the same time If you see beer, you will have a high probability of buying it, thereby increasing the sales of beer.
This association model associates different commodities by studying user consumption data, and mines the relationship between the two. When performing product correlation analysis, three indicators can be used to judge the relationship between products, and these three indicators are support, confidence and promotion. Support refers to the probability of purchasing two products at the same time, confidence refers to the probability of purchasing another product after purchasing one product, and lift refers to the degree of association between two products, that is, whether buying a product will affect The sales of another commodity are positively affected. Through commodity association analysis, merchants can better understand consumers' purchasing behavior, so as to formulate more effective marketing strategies.
08 KANO model
Enterprises often encounter a lot of product requirements, the development team is very busy, and users seem to want everything. With limited product development resources, how can we find real user needs and give high priority to really important needs?
KANO model: It is a useful tool for classifying and prioritizing user needs. It is based on analyzing the impact of user needs on user satisfaction, and reflects the nonlinear relationship between product performance and user satisfaction.
In the KANO model, according to the relationship between different types of needs and user satisfaction, the factors that affect user satisfaction can be divided into five categories: basic needs, expected needs, exciting needs, indifferent needs, and reverse needs. Type needs.
The KANO model can help us meet business needs well. Starting from the two dimensions of availability and satisfaction, we can distinguish and sort the new functions in CRM, so as to know: which functions must be available, otherwise they will be lost. Directly affect user experience (basic attributes, expected attributes); which functions will not cause negative impacts if they are not available, and will bring surprises to users if they are available (exciting attributes); which functions are optional, whether they are available or not It will not have a big impact on users (no difference factor).
Obtaining user needs is carried out through questionnaires. The KANO questionnaire consists of two questions, positive and negative, for each quality characteristic, respectively measuring the user's response to the presence or absence of a certain quality characteristic.
When the answer to the positive question is "I like it" and the answer to the negative question is "I don't like it", then in the KANO evaluation form, this functional characteristic is "O", that is, the expected type.
If the answer to the user's positive and negative questions is combined, it is "M" or "A", then the function is classified as a basic need or an exciting need. Among them, R means that the user does not need this function, or even has an aversion to the function; I means that there is no difference in demand, and the user does not care about this function. Q indicates a questionable result, which generally does not appear (unless the question is asked in an unreasonable way, or the user does not understand the question well, or the user makes a mistake when filling in the answer to the question).
Simply put:
- A: excited type;
- O: expected type;
- M: essential type;
- I: No difference type;
- R: reverse type;
- Q: Questionable results.
Note: The above comparison table is only the most common classification method. In practice, it can vary from person to person, product, company, and region (especially regarding the definition of "R" and "O"), because satisfaction itself is difficult to measure.
It is not difficult to see from the above table that the function "Dial a phone call" in the address book may have scores in six dimensions. After adding the proportions of the same dimension, the sum of the proportions of the six attribute dimensions can be obtained. The sum The largest attribute dimension is the attribute attribute of the function.
It can be seen that "providing the "call" function in the address book" is an exciting demand. It means that without this function, users will not have strong negative emotions, but with this function, users will feel satisfied and surprised.
If you only judge this one requirement, then you can stop here at this step. If it involves the sorting and grading of multiple requirements, you need to go to the next step— calculate the better-worse coefficient
Better, which can be interpreted as an increased satisfaction coefficient. The value of Better is usually positive, which means that if the product provides a certain function or service, user satisfaction will increase. The larger the positive value/closer to 1, the stronger the effect of improving user satisfaction and the faster the satisfaction rises.
Worse, can be called the unsatisfactory coefficient after elimination. The value of Worse is usually negative, which means that if the product does not provide a certain function or service, user satisfaction will decrease. The larger the negative value/closer to -1, the greater the impact on user dissatisfaction, the stronger the impact on satisfaction reduction, and the faster the decline.
Therefore, according to the better-worse coefficient, the projects with higher absolute scores of the two coefficients should be implemented first.
Its calculation formula is as follows:
- Increased satisfaction factor Better/SI=(A+O)/(A+O+M+I)
- Dissatisfaction coefficient after elimination Worse/DSI= -1 * (O+M)/(A+O+M+I)
Suppose a product wants to optimize 5 functions, but it is not sure which functions are most needed by users. By calculating the better-worse coefficient values of these 5 functions and drawing them as a scatter plot, the graph can be divided into four quadrants to determine the priority of each function.
- Quadrant 1: Also known as Expected Factors. In this quadrant, the value of the better coefficient is high, and the absolute value of the worse coefficient is also high, which means that these factors have a greater impact on user satisfaction. For example, feature 2 belongs to this quadrant. If the product provides this feature, user satisfaction will increase; conversely, if this feature is not provided, user satisfaction will decrease.
- The second quadrant: excitatory factors. In this quadrant, the value of the better coefficient is high and the absolute value of the worse coefficient is low, which means that these factors have a great effect on improving user satisfaction. For example, feature 1 belongs to this quadrant. If the product does not provide this feature, user satisfaction will not decrease; but if this feature is provided, user satisfaction will be greatly improved.
- The third quadrant: non-differential factors. The low value of the better coefficient and the low absolute value of the worse coefficient mean that these factors have no significant impact on user satisfaction. For example, function 3 belongs to this quadrant. No matter whether the product provides this function or not, user satisfaction will not change much. These function points are not the focus of users' attention.
- The fourth quadrant: must-have factors. In this quadrant, the factors that fall into it are called the basic requirements, which are the functions that the product must have. For example, function 4 belongs to this quadrant. If the product provides this function, user satisfaction will not be greatly improved; but if this function is not provided, user satisfaction will be greatly reduced, indicating that this function is one of the most basic functions. one.
In a real project:
- We must first go all out to meet the most basic needs of users, that is, the must-have factors represented by the fourth quadrant. These needs are things that users think we are obliged to do.
- After meeting the most basic needs, try to meet the expected needs of users, that is, the expected factors represented by the first quadrant, which is a competitive factor of quality. Provide additional services or product features that users love, make their products and services superior to and different from competitors, and guide users to strengthen their good impression of this product.
- Finally, strive to realize the exciting needs of users, that is, the exciting factors represented by the second quadrant, and enhance the loyalty of users.
Therefore, in this example, the value of the better-worse coefficient calculated according to the KANO model indicates that the product satisfies functions 5 and 4 first, then optimizes function 2, and finally satisfies function 1.
However, it does not matter whether the user has or does not have function 3. It is an undifferentiated demand, and there is no need to spend a lot of effort to realize it.
Original link:
Extended article:
For more exciting cases and solutions, please visit the Runwise Innovation Community .