According to the American Institute Arthur Hughes database marketing research, customer database, there are three elements of magic, these three elements constitute the best indicators of data analysis:
- Recent consumption (Recency)
- Consumption frequency (Frequency)
- The amount of consumption (Monetary)
RFM analysis originally used for traditional marketing, retailing and other fields, has a variety of consumer goods or for fast moving consumer goods industry, as long as any consumer data records can be used for analysis. Then for e-commerce web sites, web site database records detailed transaction information can also be used RFM model for data analysis, especially for those who have established a customer relationship management (CRM) system sites, the results of its analysis will be more meaningful.
Explain the basic concepts
RFM model is a measure of customer value and customer Profitability is an important tool and means. RFM analysis model composed mainly by three indicators, the following definitions of these three indicators and the role of doing the simple explanation:
Recent consumption (Recency)
Consumption means the last time the user last purchase, in theory, the last spending more time close to the customer should be good customers for the goods or services provided are also most likely to have reactions. Because the definition of recent consumption indicators is a time period, and with the current time-dependent, is always changing. Recent consumption is an important indicator of marketing involves attracting customers, retain customers, and win customer loyalty.
Consumption frequency (Frequency)
Frequency of consumption is the customer number in a certain period of time of consumption. Most frequently purchased consumer loyalty also the highest increase in the number of customers to buy means to steal market share from competitors and from other people's hands to earn revenue.
According to this index, we again divided into five equal portions customers, the quintile analysis is equivalent to a "loyalty ladder" (loyalty ladder), the trick is to allow consumers to have to climb down the ladder, the sale thought of as is to push up the two customers into buying customers to buy three times, the two-time buyers into the.
The amount of consumption (Monetary)
Consumption is the amount of e-commerce site production the most direct measure, you can verify the "Pareto law" (Pareto's Law) - 80% of its revenue comes from 20% of the customers.
Data acquisition and analysis
In extract relevant data from the database before, you first need to determine the time span of data, according to the website selling different items, determine the appropriate time span. If the business is fast moving consumer goods, such as daily necessities, you can determine the time span of a quarter or a month; if sales of the product turnover time is relatively long some, such as electronic products, you can determine the time span of one year, six months or a quarter. In determining the time span after which you can extract the data within the corresponding time interval, in which:
Recent consumption (Recency), taken out of the data is a point in time, need for the current point in time - most recently spending time as the measure of value, pay attention to the choice of units and unity, regardless of the hours, days, respectively;
Consumption frequency (Frequency), this indicator can be directly in the database COUNT the number of consumer users get;
The amount of consumption (Monetary), each customer can be the amount of the sum of all consumption (SUM) obtained.
For three indicators of the data, the need to calculate the mean data for each indicator, respectively, AVG (R), AVG (F), AVG (M) expressed by each customer's last three indicators compared with the mean , customers can be broken down into eight categories:
| Recency | Frequency | Monetary | Customer Type |
| ↑ | ↑ | ↑ | Key-value customers |
| ↑ | ↓ | ↑ | Important to develop customer |
| ↓ | ↑ | ↑ | Important to keep customers |
| ↓ | ↓ | ↑ | Important to retain customers |
| ↑ | ↑ | ↓ | General value customers |
| ↑ | ↓ | ↓ | General development of the customer |
| ↓ | ↑ | ↓ | Generally keep customers |
| ↓ | ↓ | ↓ | Generally retain customers |
- "↑" indicates greater than average, "↓" indicates less than the mean, thank nancy reminder, table Recency arrow should be reversed, and the following diagram is
Results show
RFM model consists of three indicators, coordinate plane graph can not show, so three-dimensional coordinate system used here show, an X-axis represents Recency, Y-axis Frequency, Z-axis Monetary, coordinates of the eight quadrants, respectively, 8 class of users, according to the classification of the table, you can use the following graphic description:
RFM analysis there are some defects, it can only analyze a transaction's users, but not on the visited Web site users as indicators of consumption restrictions can not be analyzed, so that potential customers can not be found. Therefore, the analysis of e-commerce website users, the richness of the data because the site - only to have trading data, and can collect the user's browser to access the data, can be extended to a broader perspective to observe the user, this quantitative analysis will In the following website for detailed analysis of the user.
Over part of the concept comes from the MBA Baike think tank
>> In this paper, BY-NC-SA agreement, please indicate the source: Website Data Analysis >> "e-commerce website RFM Analysis"
Related Articles:
Recent consumption (Recency)
Consumption frequency (Frequency)
The amount of consumption (Monetary)
Not limited to these three indicators can be adjusted according to their own business.
@ laird.zheng : Yes, to do data analysis must be combined with site specific business model and operational characteristics.
Can not believe that, RFM is the user a comprehensive assessment of the value of simple version? What is the difference of these two methods the use or condition?
@ zhilavie : ah, RFM originally used in traditional marketing industry, so that only the customer transaction data, and e-commerce site data can be more comprehensive, so the index can also be extended to do, is not limited to RFM.
Understand, RFM earlier generation, a different data base, thank you!
You give the customer segmentation for eight classes, I think the first category should be
Recency Frequency Monetary customers significant value type ↑ ↑ ↑ ↓ ↑ ↑ customer why not of significant value customers
@ Nancy : Thank you to remind, that Recency is as small as possible, the text does backwards, then do not check the clear, I note under the table below.
Oh, all right, let me introduce you to learn a lot.
Very good ~ ~ ~ good idea ~ ~ customer analytics bloggers have to learn the methods of analysis of potential customers? '
@ Vincent : blog analysis of all the users can refer to related articles website users to analyze this directory.
RFM uses too broad, the specific use of this model, when you really want to combine some of the specific characteristics of the business, for example, different categories of users Recency Frequency Monetary there is a greater difference, this time we must conduct RFM analysis according to different categories.
@ fly321283 : ah, combined with practical application to work.
Theory is very simple, but I do not know how to operate and to write reports.