Welcome back to our series on attribution in digital advertising. Last time around we talked about more complex forms of attribution and how to implement them. Today we'll talk about an alternative approach to attribution and how you can measure incremental value on your campaigns.
What is Incremental Measurement?
Simply put, incrementality is measuring the impact of a channel or tactic on your brand's revenue (or other key metric, but this usually comes up in the context of revenue), specifically in terms of "what would I lose if I took it away?" We mentioned Markov models in our Attribution 201 article and those are built around estimating this, but you can take that a step further and measure it directly.
The easiest way to measure incrementality is to just remove channels and look at the impact. What happens to my site visits and revenue if I turn off paid search? display advertising? social? Look at the difference in revenue or interaction over some period of time and boom, that's your impact. Of course, this carries a pretty big risk: You have to risk losing revenue for a period of time just to understand the value of a channel, and in the process lose the value that channel was providing.
This approach is easiest when it's not subtractive, i.e. when brands are starting out, they can build their marketing presence one channel at a time, and in the process easily see the incremental value of each new channel. Though this won't help them understand the additive value of a channel, nor clearly understand the value of increased spend in one channel and how those channels affect each other. The good news is, there are some other ways to measure this.
Control and Exposed Methodology
The most common way to measure incremental lift is using a control and exposed methodology. In this system, the brand's target audience is split into an "exposed" group who will be shown ads on a given platform, and a "control" group of people in the same target audience who will not be shown ads. Then performance is tracked via standard conversion and attribution methods for each, and the lift in performance - if there is any - between the exposed and control groups is basically the incremental value of that channel or tactic.
Splitting your audience between control and exposed groups can be done in a number of different ways. Some DSPs, such as Google DV360, have this functionality built-in, allowing you to create segments before your campaign launches to track control and exposed within the same audience group. If you don't have this kind of splitting tech on hand however you can accomplish the same task by serving PSAs to a portion of your target audience, pixeling those ads, and then being sure to anti-target them with future brand ads, creating a de facto control group from the same audience. The key here is ensuring that both your control and exposed groups come from the same audience, and that your tactics for reaching them are the same, i.e. you are paying the same amount to reach them in the same places. This will ensure you are not biasing results.
Once you've done that, you can compare the performance of each group, determine the lift, and you have your baseline.
Test and Learn
Setting up incremental measurement for your marketing initiatives isn't the end of the battle - it's the start of an always-on process of testing and learning. As your business evolves, you'll want to continually look at and re-measure the impact of channels and strategies, testing new channels, and looking for both audience and sales overlap between them. You don't have to periodically turn channels off necessarily, but you do need to revisit them from time to time to make sure they're still generating value. That process itself is an ongoing one you can develop, learning over time and improving.
Interested in talking about or implementing a better attribution strategy? Drop us a note in our Contact form and let us know!
Comments