Predicting and analyzing Customer Lifetime Value (CLTV).
Predicting Customer Lifetime Value (CLTV) allows you to efficiently identify high-value customers and understand their behavior. Eventually, you can effectively optimize your day-to-day marketing activities without missing important highly-engaged customers.
For online retailers, the Treasure Boxes for CLTV Prediction shows how to create a CLTV prediction model based on customer's order histories.
Input: Order histories
This Treasure Box guides helps you prepare the following transactional data based on a public dataset:
InvoiceNostring |
InvoiceDatestring |
CustomerIDlong |
Countrystring |
StockCodestring |
Descriptionstring |
UnitPricedouble |
Quantitylong |
---|---|---|---|---|---|---|---|
536365 | 2010/12/01 8:26 | 17850 | United Kingdom | 85123A | WHITE HANGING HEART T-LIGHT HOLDER | 2.55 | 6 |
536365 | 2010/12/01 8:26 | 17850 | United Kingdom | 71053 | WHITE METAL LANTERN | 3.39 | 6 |
536365 | 2010/12/01 8:26 | 17850 | United Kingdom | 84406B | CREAM CUPID HEARTS COAT HANGER | 2.75 | 8 |
... | ... | ... | ... | ... | ... | ... | ... |
C536391 | 2010/12/01 10:24 | 17548 | United Kingdom | 22556 | PLASTERS IN TIN CIRCUS PARADE | 1.65 | -12 |
... | ... | ... | ... | ... | ... | ... | ... |
Calculating CLTV and Derived Attributes
In the middle of the workflow, the data is aggregated, CLTV is calculated for each CustomerID, and attributes are enriched with basic statistics:
customeridlong |
cltvdouble |
countrystring |
recencylong |
avg_basket_valuedouble |
avg_basket_sizelong |
cnt_returnslong |
has_returnedboolean |
---|---|---|---|---|---|---|---|
14001 | 2030.33 | United Kingdom | 327 | 681.0266666666666 | 320.6666666666667 | 1 | 1 |
14911 | 132572.6200000001 | EIRE | 372 | 715.5475621890546 | 400.5721393034826 | 47 | 1 |
18144 | 2888.7499999999995 | United Kingdom | 365 | 240.72916666666666 | 111.91666666666667 | 0 | 0 |
... | ... | ... | ... | ... | ... | ... | ... |
Output: Predicted CLTV
The output of this Treasure Box is a table that contains the predicted CLTV for a subset of customers:
customeridlong |
predicted_cltvdouble |
---|---|
12355 | 68.84149528775471 |
12356 | 172.29478943269208 |
12357 | 151.84542283221555 |
... | ... |
Use-Cases
- Sales
- Marketing
- Customer Success
- Customer Support