Most retailers are stuck in a rut of measuring a channel’s success based on last-click attribution, even though the model ignores the many other touchpoints customers may hit before converting. Marketers should instead steer toward an incremental revenue model that factors in the multi-channel marketing landscape.
Why the last-click attribution model fails
The last-click attribution model, while widely adopted and implemented by default by many website analytics tools, is the worst way to measure success because it allocates 100 percent of the credit for the sale to the last channel a customer clicked on. If customers only interacted with one touchpoint, that model would be fine, but with more than 50 percent of customer interactions involving multiple channels, the last-click attribution model is ignoring key channels which influence shopping behavior along the way.
When measuring the ROI of email, this model is especially problematic since the call to action may not drive immediate conversion. Imagine this common scenario: a customer gets an email from Macy’s with a 30 percent off coupon, clicks the link in the email to browse the site and finds a new waffle iron to add to his cart. When he gets to the checkout page, he sees an empty coupon box which prompts him to search for a coupon through Google, and lo and behold he finds one on RetailMeNot. He attaches the coupon and makes the purchase.
While the purchase was originally driven by the promotional email, the last-click attribution model gives all of the credit to the affiliate site, RetailMeNot, and doesn’t recognize the importance of the email sent by the retailer.
But the flip side of last-click attribution — first-click — doesn't fare any better. As Avinash Kaushik says, “First-click attribution is akin to giving my first girlfriend 100 percent of the credit for me marrying my wife.” Any single-click attribution model is problematic because it ignores the nature of cross-channel messages.
What about fractional attribution?
Some retailers are trying to tackle the complexities of multiple cross-channel touchpoints with fractional attribution methods. Adopting a “peanut butter spread” approach where all channels that were touched receive the same amount of attribution assume too much simplicity. Other methods that use business rules to allocate a different share of attribution to the first touch, middle touches, and last touch channels are flawed because they use arbitrary rules for divvying out the share of “credit” for driving the sale.
The best attribution model: Incremental revenue
To determine if a channel is successful, a retailer should rely on an incremental revenue model, which looks at customer conversion based on the treatment they received. In a nutshell, this is done by separating a small subsection of customers who don’t receive a certain marketing message and then measuring whether there is a difference, good or bad, in the customers who received the messages (the treated group) and those who didn’t (the holdout group).
This model is easiest to implement with direct marketing channels, such as email, SMS and display, and more difficult with search and other broadcast marketing. For example, if a company chooses to not send an email to 5 percent of population of its email subscribers, it can calculate the relative ratio of the customers in the holdout group who made a purchase and customers in the treated group who received the email and also made a purchase. The incremental revenue measurement answers the question of whether the company saw an increase in the revenue per subscriber because the email was sent.
Smart marketers are realizing the potential of comparing a treated and untreated group to decide if messages are effective at driving incremental revenue online and in stores. Marketers are constantly asked to prove the ROI of their messages, and incremental revenue is perhaps the most effective model to measure the effectiveness of your relationship marketing efforts.