# AnalyticBridge

Subscribe to Vincent Granville's Weekly Digest:

# Calculating Direct Marketing ROI Analysis Using Control Groups

Time series analysis and a control group may be used to evaluate the success of a direct marketing campaign whether it is delivered via Canada Post, or is an e-mail or search engine marketing campaign. A control group is a random sample of customers who do not receive a direct marketing offer and their behaviour may be compared to a similar, randomly generated group of customers who don’t receive one. This type of direct marketing campaign analysis is performed over time to compare two groups before, during and after a specific campaign period--whether days, weeks, months, or otherwise. Observing the behaviour of a control group versus a similar group of targeted customers allows one to calculate the incremental campaign revenue generated by, and the overall return on investment (ROI), or a direct marketing campaign.

As an example, let’s imagine that there is an e-mail database of 5,000 customers, 500 of whom (TARGET_GROUP) have been randomly targeted with a direct marketing offer. Each TARGET_GROUP customer has been offered a free pen for every time, within a one-week campaign period, that he or she buys a pad of paper via our website.

Before a direct marketing campaign analysis may be conducted, it is important to know the following: both our pen and pad of paper have long been on the market and the same offer is being promoted on TV during the e-mail campaign.

A time series, control group e-mail marketing campaign analysis will seek to answer the following questions:

1. How much more revenue was generated from pen and paper sales by the e-mail campaign?

2. Given the costs, what is the overall return on investment (ROI) for the e-mail marketing campaign?

TARGET_GROUP’s pen and paper buying behaviour may be compared to a randomly generated CONTROL_GROUP of 500 customers who receive no offer. Both groups are best evaluated on a week-to-week basis: before, during and after the campaign period. A \$PRODUCT variable will capture the amount of pen and paper revenue that is generated from each TARGET_GROUP and CONTROL_GROUP customer.

Graphing the TARGET_GROUP and CONTROL_GROUP customers, over time, according to the \$PRODUCT variable will provide an answer to question 1: how much revenue did the e-mail campaign generate. If, during the one week campaign period, TARGET_GROUP customers spent \$100 more on pen and papers than CONTROL_GROUP customers, the e-mail campaign generated an incremental \$100 in revenue.

Determining an overall ROI for the campaign requires that the costs involved to create and run it be subtracted from the \$100 in revenue generated by it.

Let’s assume that the following costs were incurred in order to execute the e-mail marketing campaign:

Creating and testing the e-mail blast template - \$5

Programming the website to accept online payments - \$25

Staff time to plan and manage the e-mail campaign - \$30

Since our total costs were \$60, an analysis would reveal that the overall ROI for the e-mail campaign is \$40.

This example provides a simple illustration of how a time series analysis using control groups may be used to evaluate the success of a direct marketing campaign.

Views: 1099

Comment

Join AnalyticBridge

Comment by Rajesh Viswanath on September 24, 2010 at 11:06am
This is a great article, and it does demonstrate the application of time series models for test control studies. When evaluating the impact of some event on test versus a control, if weekly change is not of interest, it is possible to evaluate the impact of a certain event using Analysis of covariance methodology. ANCOVA will tease out the effect of other nuisance factors.

1

2

3

4

5

6

7

8

9

10