Why Mobile Marketers Need To Think Like Statisticians

A statistician's wife has twins. He is delighted. He rings the minister, who says, "Bring them to church Sunday and we'll baptize them.” "No," replies the statistician. "Baptize one. We’ll keep the other as a control.”

Just a little statistician humor for you! But jokes aside, mobile marketers need to embrace statistical models to avoid drawing inaccurate conclusions about their marketing, specifically when retargeting active users. Too often, marketers use return on ad spend (ROAS) to assess the effectiveness of their retargeting campaigns, but this can be a misleading figure because it does not account for actions that might have happened anyway, even if you had not served users ads.

When scientists conduct clinical research, they commonly use the intention-to-treat (ITT) protocol to provide an unbiased comparison of the treatment groups. The ITT approach is also the key to calculating incrementality  the measure of revenue lift generated by your ad spend, but the majority of marketers aren’t using it. Here’s a simplified look at how it works.

1.     Divide users into a control group and a target group.
   a.     Control group: Users who will not be retargeted
   b.     Target group: Users who are allowed to be retargeted but not everyone will see an ad due to budgetary constraints and users’ RTB visibility  



2.     After enough impressions have been delivered, compare the entire target group KPI (e.g., revenue or app-open) with the entire control group KPI. The volume of impressions you need to serve to be statistically significant depends on the difference in performance between the two groups. That’s another example of when and why marketing departments, and the vendors they work with, need to understand statistics and avoid oversimplifying.

Let’s compare an incrementality analysis to an ROAS analysis to illustrate the need for more sophisticated calculations.

You pay $50K for a retargeting campaign. Your retargeting partner creates a control group and a target Group. Over this time period, your control group spends $30K and your target group spends $70K. To calculate incrementality:

$70K (target group spend) - $30K (control group spend) = $40K
$40K (difference between target and control group spend) - $50K (your ad spend) = $-10K (the incremental value—or in this case, the lack thereof—of your retargeting campaign.)

Now imagine you had used ROAS to analyze your campaign instead. Spending $50K to generate $70K in spend sounds great, right? You would probably scale. This is why we need mathematics, people.

The majority of marketers are mindful of the risks associated with targeting active users. To avoid paying for conversations that would have happened anyway, many avoid retargeting active users all together. This is a huge missed opportunity, particularly in the mobile app space, where the majority of app revenue is generated by a small percentage of users. Using highly targeted ads to drive more action from your existing users—the ones who are most likely to take revenue-generating action—is a cost-effective marketing tactic. Incrementality makes it possible.

This year will mark the first-time mobile ad spend surpasses budgets allocated for TV, according to eMarketer. The reason advertisers embrace mobile is, in part, because everything is measurable. You can target and engage your audience with unprecedented precision—but not without the right methods and tools. To truly master measurement, you’ve got to put your statistician hat on.
Next story loading loading..