CMO One-to-One: The Search for a Better Attribution Model
No easy solutions
Author: Maria Minsker
September 1, 2017
Online retailer Boxed may be a young, digitally native brand, but the company still struggles with attribution. In fact, the brand is in the middle of an attribution overhaul as it works to find the best model for understanding how well its various advertising and marketing channels are working. Jackson Jeyanayagam, CMO of Boxed, spoke with eMarketer’s Maria Minsker about his vision for an attribution solution.
eMarketer: What models are you currently using for attribution?
Jackson Jeyanayagam: We're actually in the process of reassessing how to appropriately measure all of our channels. We’re trying to figure out the best tool and best approach, given that we have so many channels driving conversion.
In the past, we relied on last-touch and last-click attribution, but that didn’t serve as a good indicator of what was working best for us. For instance, we rely heavily on search, Facebook, display and programmatic advertising, but it's difficult to pinpoint which channels are working the hardest.
eMarketer: How close are you to finding a better attribution model for your brand?
Jeyanayagam: We’re moving toward it. Digital attribution is a little easier to understand, but we’re still trying to figure out the appropriate way to measure attribution above the line. We know, for example, that we don't want to use a media mix model because it requires too many assumptions. Other than that, we've vetted a lot of tools, but we want to take a step back and think about the full funnel before making any decisions.
eMarketer: Why wouldn’t a media mix model be a fit for your brand?
Jeyanayagam: A media mix model makes sense for companies when they don't have a direct-to-consumer touch, or when they have a lot of above-the-line, big brand-building efforts in place. In those cases, they need to take into consideration what's running and when, what impressions they’re getting and how reach is performing. We're not spending enough on above-the-line channels to make full use of a media mix model.
eMarketer: What are your key goals as you rebuild your attribution model?
Jeyanayagam: We’d like to know which channels are working hardest for us as well as how to scale up channels that we’re not using enough. We also want to know how to prioritize partners across those different channels. For example, we work closely with Pepsi, but fewer people search the internet for different types of soda than baby products. There’s a higher decision tree involved with the latter, so I might invest more in search advertising for our P&G partnership than our Pepsi partnership.
“We’d like to know which channels are working hardest for us as well as how to scale up channels that we’re not using enough.”
eMarketer: How important is it for you to have offline data integrated with online data?
Jeyanayagam: It’s very important, and it’s the gorilla in the room. No one has solved how to integrate offline data, unless there’s a promo code attached to an out-of-home (OOH) ad or a TV spot. The challenge is properly assessing when someone saw that billboard or saw the TV ad, and then came back and converted as a result. It’s an industry issue. But there are some solutions out there that track device IDs and monitor how quickly consumers take actions after being exposed to something, so that’s promising.