We live in a world of “more.” We want to buy more, experience more things, and have more choices. This “more” is incremental because it highlights an increase over what we usually expect in our day-to-day lives.
As a parent, incrementality is a part of my motivational toolkit—the reward I provide to convince my kids to do something they don’t necessarily want to do. My son—ever the negotiator—always finds an angle to get more. Every morning, I have the same debate with him that I always do, telling him that if he finished his homework, he could have an extra hour on his iPad. Since he normally gets one hour standard, offering him an extra hour is the incremental time he would earn by doing his homework. By increasing his total iPad playtime from one to two hours, he is rewarded with a 2x increase over the standard baseline, or 2x incremental gains.
In the world of advertising, marketers want to sell us more stuff, and this is where the digital advertising term “incrementality” comes in. When someone says they are trying to measure incrementality, they are attempting to answer the age-old question: Did my marketing dollars translate to selling more or would those consumers have purchased anyway?
Easy to get click happy
As I’ve explained in previous blog posts, clicks are not a strong measurement for display media. More and more, I am getting asked the questions:
- How do I know if my programmatic media is working without using click measurement?
- How do I know I received added sales or leads from the ads that I am running on programmatic media?
- How can I measure the incremental views that lead to a conversion that was specifically due to working with a particular partner?
Incrementality is the solution that marketers use to answer this question. While it is not a new solution, it is one that many marketers don’t understand in terms of how it works and how we can set it up correctly to enable cleaner results.
Let’s help you sound smarter
Now that you understand what incrementality is and why clicks aren’t a good proxy, let’s talk about how this type of measurement is set up. Let’s start with the definition:
Incrementality Test: A type of A/B test that has 2 balanced groups, one who receives a branded message and the other group that doesn’t receive a branded message.
The idea is that if your two groups are statistically similar, then we can make a comparison across the groups and attribute the change in the exposed group to the advertising.
What are the goals aligned with an incrementality test?
- Validate display ad impact compared to organic conversions that would have happened anyway
- Validate a single vendor audience tactic
- Compare incrementality across programs
In setting up an incrementality test, there are three main types:
- Post Hoc Incremental Test is a test where viewable impressions are compared with the non-viewable impressions and we can approximate the lift on conversion based on this comparison. This is most often the first step to understanding directionally if the campaign is showing strong positive lift. Often this test overestimates the amount of lift, but we can use it as a directional indicator if the campaign is performing. This type of analysis is possible after a campaign runs and requires no setup.
- Non-PSA/ Holdout Test is a test where we set up a test audience who will see branded messaging and a holdout / control audience who will not see any ads associated with the brand, and the two groups are compared to calculate a lift. Through this process, we randomly select a percent of all the cookies that are currently in existence or will be added during the campaign period to be in each grouping. Standard control sizes range from 10% to 50%, depending on a variety of factors including campaign sizes, time campaign will be running, and number of average conversions that take place in a particular time period.
- A PSA Test is seen as the most scientific of incrementality tests. In this test, the test audience sees branded messaging and the control audience receives a Public Service Announcement (PSA) ad (e.g., a Smokey the Bear type ad), and the results between the two groups are compared. Similar to a Non-PSA/holdout, standard control sizes range from 10% to 50%, depending on a variety of factors including campaign sizes, time campaign will be running, and number of average conversions that take place in a particular time period. In addition to balancing the pre-campaign experience, a PSA takes into account the in-campaign attributes of a person such as the quality of an ad placement, the type of site visited and the frequency an ad is received among other things.
Across any of the incrementality solutions, there is a consistent formula one can follow to calculate the incremental lift:
So if we go back to the artful negotiation I have every morning, this is the math I’m doing in my head:
And the answer is….+100% incremental lift. In a world where my son wants to buy more, experience more things, and have more choices, those twofold incremental iPad time gains are his friend. And for marketers trying to find and convert new customers, incrementality is your friend too, rewarding your bottom line and allowing you to measure “more.”
If you missed my earlier two-part blog series on being “ruled by the click,” read You Can’t Measure Display with a Yardstick or Clicks and Phishing for Clicks. You can also check out The Demographics of Clicking. To understand how to implement incrementality testing as a measurement tool for your next campaign, reach out to your account manager or contact us.