14 Feb Leverage Uplift Modeling To Optimize ROI For Direct Mail Campaigns
Most marketers know that direct mail touts an impressive response rate between 5% and 9%.
In fact, direct mail holds a considerable advantage over email, social, display, and paid search- which all have a response rate of up to 1%. However, the flip side of the direct mail coin reveals a lofty CPM.
This high cost can quickly negate direct mail’s favorable response rate over alternative channels. Marketers can’t ignore how expensive direct mail can be, especially when comparing ROI by channel.
Just as Spiderman’s Uncle Ben teaches us to balance power and responsibility, marketers need to balance direct mail response rates with higher CPMs.
So what can you do to ensure your direct mail campaigns maintain their lift while reducing the cost of deployment? Utilize “Uplift Modeling” to assign a lift score at the individual level.
What Is Lift?
Incrementality or Lift is the effect on sales after receiving a marketing stimulus:
– A positive lift score indicates an increase in sales as a result of your campaign.
– A zero score indicates that your campaign had no impact on overall sales.
– A negative lift score indicates your campaign adversely affected overall sales.
At the campaign level, measuring and understanding lift is the best way to justify whether or not you should continue executing a particular campaign.
At the individual level, “Uplift Modeling” can be used to understand whether or not you should target a specific individual with the campaign in question.
From there, you can optimize Direct Mail campaigns by targeting only the individuals who are more likely to convert after receiving a touch. The end result is equal or increased campaign lift with substantially reduced spend, and ultimately improving ROI.
Let’s walk through how you can assign an Uplift value at the individual level with an example. Our team of Data Scientists uses a combination of two models, which creates the output we’ve illustrated in the matrix below:
First, each individual runs through Model 1 which outputs a score measuring their likelihood to convert if they do not receive direct mail. Then, each individual runs through Model 2, which scores their likelihood to convert if they do receive direct mail.
The combination of the 2 model scores identifies if an individual is more, less, or equally likely to convert if they are targeted by a direct mail campaign.
Note that the diagram above includes three groups that should be excluded from future campaigns: “Do Not Disturb”, “Lost Cause”, and “Sure Thing”.
– The Do Not Disturb group includes those who are more likely NOT to convert if you send them a mail piece.
– The Lost Cause group includes those who will not convert regardless of whether they receive a mail piece.
– The Sure Thing group includes are those who WILL convert whether they receive the mail piece or not.
The Bottom Line
All three groups represent wasted spend within your campaign. Reduce spend on individuals in these groups, and focus on the individuals that fall into the “persuadable” category instead. This includes those individuals who respond positively to an additional direct mail touch, and they should yield an equivalent overall campaign conversion rate with reduced spend, and ultimately boosting ROI for your campaign.
If you would like to further discuss your direct mail strategy, execution, or the use of modeling to improve campaign ROI send us an email at firstname.lastname@example.org. Make sure to visit our blog for weekly updates on trending industry topics from our team of experts in digital marketing strategy.