At this point, you’ve probably seen headlines like “Last click attribution is dead.” But despite an increasing share of the industry looking to move away from a last click attribution, the majority of marketers still evaluate campaign success with this approach.
Think about it this way: assigning full credit to the last marketing interaction is the equivalent of going to the movies and assigning 100% credit to the poster outside the theatre advertising the film you already planned to see.
Not only is it inaccurate, it’s actually hurting your business. That’s because how you measure success dictates how your vendors are optimizing—and the dirty little secret of last click conversion-based optimization is that it’s built to target people who would likely convert on their own.
Enter incrementality. By removing that touch point for a subset of users, and then seeing what percentage of them purchase anyway, it answers the question, “Do my ads drive actual value, or just claim credit for an action that would naturally occur?” This allows us to determine if a given campaign caused a conversion, or just touched one. Knowing this ensures that dollars are directed to the former, removing wasted ad spend, and optimizing towards actually driving net new sales.
Conversion optimization sounds great on paper. Of course, we want to optimize for purchases. However, what these algorithms do is predict the likelihood that someone is going to make a purchase. Then, they make sure that they reach those people. In other words, they are aggressively going after people that are the most likely to convert.
Think about your own experience with retargeting. When are you retargeted most aggressively? I’d wager it’s immediately after you put something in an online shopping cart, but don’t purchase right away. If you hesitate and happen to check Facebook or read an article, you can expect to see page takeovers with ads for the product that you’re very likely to buy.
Now think about this, if you had to predict which group would convert at a higher rate, independent of marketing, which would you pick:
People who recently put something in the shopping cart and still have an active session on your site.
People who viewed 3 products 2 weeks ago but haven’t been back to your site since.
If your answer is the first group, then why is your ad spend concentrated on the people most likely to purchase on their own? It’s because vendors know that you’re measuring based on last click. They’re scrambling to get a touch point between that act of putting something in the shopping cart and the actual purchase event. You’re paying that vendor for conversions that would happen on their own, and that's why last click attribution is dead.
Incrementality Optimization (iCPA and iROAS)
In contrast, incrementality-based optimization predicts the likelihood that someone will convert as a result of seeing an ad. To do so, the engine needs to understand the causal event that drives a desired outcome. In the case of ads, this means identifying the positive, negative, or neutral impact that an ad has on a business outcome. That is accomplished by running an ongoing hold-out group and measuring the impact on key business metrics between that group and the group that is exposed to ad campaigns. This allows you to see which campaign configurations are actually causing net new sales, vs. which are not.
Connecting This to Your Marketing Efforts
What we’ve learned from running such tests is that the campaign configurations that are most common for hitting CPA targets are actually the worst at driving incremental lift across sales and revenue. Focusing on things like recency and cart abandonment allows vendors to represent that they’re driving value when in reality they’re simply taking credit.
On the flip side, tactics that look worse when measured on a last click basis, like opening up your targeting window to 30 days since site visit, or going after a broader slice of the funnel, yield significantly higher increases in lift across sales and revenue.
A real-world example of a test we ran with a large, well-known apparel brand proved exactly this. We tested a common click-to-conversion optimization strategy targeting a $50 cCPA against an incrementality-based strategy.
While the click-to-conversion strategy had no problem hitting the CPA target, it actually negatively impacted revenue for that group as CPA optimization often includes invasive advertising placements that can actually turn consumers off from purchasing, the incrementality strategy failed to hit the CPA target. But, it did actually drive $56K in revenue that would not have otherwise occurred. In other words, one tactic looked better from a CPA perspective, but the other tactic actually increased overall sales. Which speaks to the fundamental problem that many companies are actually optimizing towards something that could actually have a negative impact on their business.
The reality is every site, brand, and business model is different. And as a result, the right way to determine the ideal set up for your retargeting efforts is to run a scaled, ongoing incrementality test on various campaign configurations. Without that, there is a good chance that clinging to last click isn’t just imperfect, it’s actually hurting your business.