What problem does the business want to solve?
The client is a leading American retail company, primarily involved in the sale of office supplies and related products. It aims to
drive traffic to online and retail stores
increase customer acquisition
achieve incremental results
What are the business goals?
Predict whether a customer really need a coupon to motive the purchase
Find the probability of purchase for the persuadable customer
Model Solution
Data Universe
Prior campaign results of 13-month consumers information
17M campaign circulations in total, 6M unique customers
15 M Treatment Group
2 M Control Group
Measurement Window & Prediction Window
Campaign start to end date (40 days around)
Dependent Variable
Whether an account purchase using coupon redemption during measurement window (Comparing with the holdout with purchase)
Data Selection
Data Type
Reward/non-rewards customers
Payment types
Coupon need
Payment types
Product category
High valuable customer
Recency
Variable Selection
The top 12 drivers are shown below in order of driver importance
non-rewards purchase amount
payment method: using cash
coupon related variable: redemption rate
coupon related variable: retail sales
coupon related variable: weeks of use coupon
coupon related variable: using coupon vs total sales ratio
coupon related variable: redemption amount
payment method: combing with staples coupon
Facilities & Supplier average margin
reward customer margin
all net margin historically
most recently net margin
What’s our approach?
Based on the variables we can derive even more
How did we use SSB to generate derived variables?
Examples of the power of the new variables?
original and the new and compare them
the purpose is to showcase why FocusKPI is unique
We can automate the estimation and result in the higher lift and more power models
Bring in the tableau reporting linked to the model results
put in some screenshots of SSB
Exploratory Data Analytics
Select appropriate prior campaign data
Identify key drivers to split the universe into two groups
Validate model
Rank and profile customers by model scores
Select top decile (retail sales) for the campaign
Direct Mail (DM) Campaign Objective
Reactivate lapsed audience with a coupon offer available to redeem in-store to drive incremental results
Direct Mail (DM) Campaign Operation
Postcard coupon was sent to a total of 530K Treatment Group (320K Business and 210K Business-like Customer), with a Campaign cost of $630K and measurement time of 40 days
What are the insights?
As a result, after 40 days of measurement, both groups performed significantly better. The campaign drove significant incremental response and net sales.
WHAT PROBLEM DOES THE BUSINESS WANT TO SOLVE?
The client is a leading American retail company, primarily involved in the sale of office supplies and related products. It aims to
drive traffic to online and retail stores
increase customer acquisition
achieve incremental results
WHAT ARE THE BUSINESS GOALS?
Predict whether a customer really need a coupon to motive the purchase
Find the probability of purchase for the persuadable customer
MODEL SOLUTION
Data Universe
Prior campaign results of 13-month consumers information
17M campaign circulations in total, 6M unique customers
15 M Treatment Group
2 M Control Group
Measurement Window & Prediction Window
Campaign start to end date (40 days around)
Dependent Variable
Whether an account purchase using coupon redemption during measurement window (Comparing with the holdout with purchase)
Data Selection
Data Type
Reward/non-rewards customers
Payment types
Coupon need
Payment types
Product category
High valuable customer
Recency
Variable Selection
The top 12 drivers are shown below in order of driver importance
non-rewards purchase amount
payment method: using cash
coupon related variable: redemption rate
coupon related variable: retail sales
coupon related variable: weeks of use coupon
coupon related variable: using coupon vs total sales ratio
coupon related variable: redemption amount
payment method: combing with staples coupon
Facilities & Supplier average margin
reward customer margin
all net margin historically
most recently net margin
WHAT’S OUR APPROACH?
Based on the variables we can derive even more
How did we use SSB to generate derived variables?
Examples of the power of the new variables?
original and the new and compare them
the purpose is to showcase why FocusKPI is unique
We can automate the estimation and result in the higher lift and more power models
Bring in the tableau reporting linked to the model results
put in some screenshots of SSB
EXPLORATORY DATA ANALYTICS
Select appropriate prior campaign data
Identify key drivers to split the universe into two groups
Validate model
Rank and profile customers by model scores
Select top decile (retail sales) for the campaign
DIRECT MAIL (DM) CAMPAIGN OBJECTIVE
Reactivate lapsed audience with a coupon offer available to redeem in-store to drive incremental results
DIRECT MAIL (DM) CAMPAIGN OPERATION
Postcard coupon was sent to a total of 530K Treatment Group (320K Business and 210K Business-like Customer), with a Campaign cost of $630K and measurement time of 40 days
WHAT ARE THE INSIGHTS?
As a result, after 40 days of measurement, both groups performed significantly better. The campaign drove significant incremental response and net sales.