The original link:tecdat.cn/?p=393
Original source:Tuo End number according to the tribe public number
How to measure the value of e-commerce stock users? Was it your last purchase? Amount spent? Or the number of purchases? What model of segmentation is most effective for marketing campaigns to increase user responsiveness?
If an e-store launches a marketing campaign in April 2017, it will need to carry out coupons, SMS and email marketing to regular users. But marketing costs are only enough to support 2,000 users.
Then we can select the most likely 2000 users through the RFM model.
RFM Introduction to the
RFM is a method used to analyze customer value. Often used in database marketing and direct sales.
Indicates the meaning of RFM
Recent purchases – What has the customer recently purchased?
Frequency of purchase – How often do they purchase?
Purchase value – How much do they pay?
Most businesses will retain data about customer purchases. All you need is a table that contains the customer name, purchase date, and purchase value.
Recent purchases = Max (10 – number of months that have passed since the customer’s last purchase)
Frequency of purchases = Max (number of purchases in the past 12 months)
Purchase value = customer’s highest order value
Customer Analysis section
Returns RFM data for users of different merchants
The table name: userrfm
| The user (Userid) | Last Recency [L1] | Consumption Frequency | Amount (Monetary) | Merchants (Busid) |
|---|---|---|---|---|
| 100001 | 1 | |||
| 100002 | 1 | |||
| 100001 | 2 |
Customize the analysis section
Set the threshold of purchase times of new customers, repeat customers and old customers as P, q and R (parameters are set in R and passed in other ways later)
The table name: frequency
| Member type | Conditional Setting (F) [L2] | membership | Members of | Consumption amount | The guest unit price | Business name (Busid) |
|---|---|---|---|---|---|---|
| Interested customers | 0 | 120 | 1 | |||
| A new customer | p | 20 | 1 | |||
| Repeat customers | q | 10 | 1 | |||
| Old customers | r | 3 | 1 | |||
| Interested customers | 0 | 120 | 2 | |||
| A new customer | p | 20 | 2 | |||
| Repeat customers | q | 10 | 2 | |||
| Old customers | r | 3 | 2 |
Similarly, set the thresholds of customer patronizing days to p, Q and R respectively (parameters are set in R and passed in other ways later).
Number of days visited (R)
The threshold of attrition period is ABCD (the parameter is set in R and passed in other ways later)
The table name: Recency
| Member type | Conditional Setting (F) [L3] | membership | Members of | Consumption amount | The guest unit price | Business name (Busid) |
|---|---|---|---|---|---|---|
| Late-sale customer | 0~a | 120 | 1 | |||
| Active customers | a~b | 20 | 1 | |||
| Silent period customer | b~c | 10 | 1 | |||
| Sleeping clients | c~d | 3 | 1 | |||
| Churn customer | >d | 2 | 1 | |||
| Late-sale customer | 0~a | 2 | ||||
| Active customers | a~b | 2 | ||||
| Silent period customer | b~c | 2 | ||||
| Sleeping clients | c~d | 2 | ||||
| Churn customer | >d | 2 |
Number of days visited (R)
The threshold of attrition period is ABCD (the parameter is set in R and passed in other ways later)
The table name: Recency
| Member type | Conditional Setting (F) [L4] | membership | Members of | Consumption amount | The guest unit price | Business name (Busid) |
|---|---|---|---|---|---|---|
| Late-sale customer | 0~a | 120 | 1 | |||
| Active customers | a~b | 20 | 1 | |||
| Silent period customer | b~c | 10 | 1 | |||
| Sleeping clients | c~d | 3 | 1 | |||
| Churn customer | >d | 2 | 1 | |||
| Late-sale customer | 0~a | 2 | ||||
| Active customers | a~b | 2 | ||||
| Silent period customer | b~c | 2 | ||||
| Sleeping clients | c~d | 2 | ||||
| Churn customer | >d | 2 |
The guest unit price(M)
The threshold of attrition period is L ml m h (the parameter is set in R and passed in other ways later)
The table name: Monetary
| Member type | Conditional Setting (F) [L5] | membership | Members of | Consumption amount | The guest unit price | Business name (Busid) |
|---|---|---|---|---|---|---|
| Low value customer | 0~l | 120 | 1 | |||
| Low to medium value customers | l~ml | 20 | 1 | |||
| Mid-value customer | ml~m | 10 | 1 | |||
| Mid to high value customers | m~h | 3 | 1 | |||
| High value account | >h | 2 | 1 | |||
| Low value customer | 0~l | 2 | ||||
| Low to medium value customers | l~ml | 2 | ||||
| Mid-value customer | ml~m | 2 | ||||
| Mid to high value customers | m~h | 2 | ||||
| High value account | >h | 2 |
Model realization (R language )
Connect to mysql database
Get data from the database
The original data
Customer Analysis section
Customize the analysis section
Buy the number
Coming days
Unit Price (M)
RFM 3 d crosstab analysis
Interface:
1. Number of customers/Proportion
2. Average purchase amount
3. Cumulative purchase amount
R-value analysis (time span [0,1080]
1. Index of F value
2. M value index
3. Indicators of membership level
F value analysis (F value [1,20],(20,+info))
1. R value index
2. M value index
3. Indicators of membership level
M-value analysis (m-value interval selection, purchase amount (average purchase amount, accumulated consumption amount), 20 lines)
1. R value index
2. F value index
3. Indicators of membership level
These reports comprehensively show the various dimensions of RFM model analysis. Therefore, we can clearly analyze the relationship structure of a customer group, and push different businesses according to the actual business and different groups.