Attribution Should Be About Contribution Chains Rather Than Awarding Credit
Marketers put a lot of time and energy into determining which channel gets credit for a conversion -- but contributor Soren Ryherd believes that this approach is missing the point.
The conversation about attribution is all wrong. Even the name “attribution” conveys the wrong meaning. Much of the attribution discussion — and even the attribution models we use — focuses on dividing up the chain of touchpoints to determine which channel gets “credit.”
Yet multi-touchpoint engagement implies that the whole chain of touchpoints is what creates value — not each independent touchpoint.
Current attribution models, such as First-In, Last-In, Even Distribution, or custom weighted models, all look to discretize the value based on philosophical approaches.
For example, proponents of Last-In (if there are any… Bueller?) would argue that the last step in the journey should get the credit, as it is the closest step to value. First-In proponents argue that early steps are tied to the motivating or educational stages of engagement, and, therefore, deserve to be more highly valued.
The reality is, all of these are guesses. And we don’t have to guess. We can take the information from our data sources and tests to create attribution models based on empirical data regardless of assumptions.
It is time to shift the discussion, and our models, from dissecting touchpoint chains to “credit” individual channels, towards reinforcing the tactics that create chains of value in the first place.
The current attribution discussion stems from over a century of channel-specific marketing, and a channel-specific worldview that has carried over into our digital programs, budgets, and even operational structures.
When the evolution of tracking data began revealing user engagement paths involving multiple channels, we stuck to our channel-centric view and attempted to shoehorn multi-channel data into our single-channel reports, tools, and execution. In larger media programs, this tied back to departmental budgets and siloed operations within marketing.
But here’s the thing. Users don’t care about budget silos, departments or fiefdoms. They care about their needs.
When value is created by multiple channel touchpoints, what users are telling us is that it is the combination of channels that they need for engagement. And it is becoming increasingly apparent that attribution models that attempt to discretize value over different channels without looking at the impact on the creation and support of the entire chain may not be as efficient as they appear.
A completely different approach is to use Chain-Based Attribution.
Chain-Based Attribution looks at understanding the value of the chain by working backward from customer value, then making tactical decisions in valuing the touchpoints to maintain the entire chain.
Chain-Based Attribution is less concerned about the specific placement of the touchpoint, such as is used in First-In or Last-In models, as opposed to the fragility of the touchpoint. Protecting the chain is everything.
Most multi-touchpoint chains begin with a non-branded touch that exposes or reinforces the connection with a user and a brand. This can include non-branded search (paid or organic), display, social media, links, YouTube ads, or more.
Almost all chains end with a branded touchpoint, most commonly a brand search or a direct (No Referrer) visit, or remarketing exposure. In the middle can be a mix of almost anything, including brand and non-brand, remarketing, and more.
What makes a touchpoint or channel fragile in the touchpoint chain? Which run the risk of not being present if we don’t tactically support them?
Almost always, auction-based paid media is the most fragile. The nature of the auction is that in a competitive field, not supporting paid media to the full value possible based on the value of a customer means that the channel is at risk of not being present.
If that touchpoint is no longer present, the chain no longer exists. If the entire chain no longer exists, it cannot create value from those users that need that combination of touchpoints for engagement.
Other factors contribute to fragility in touchpoint chains. For example, non-brand organic keywords that rank high on Search Engine Result Pages (SERPs) may be at risk in a page redesign.
Display, Video, and other highly interruptive media may be at risk because the initial engagement is at an impression level with no click ever being recorded in the touchpoint stack. Therefore, we may need to develop methods (such as statistical modeling) for valuing those types of media relative to the volume of touchpoint stacks that only show branded and direct touches.
In practice, Chain-Based Attribution most closely resembles First-In models, but for radically different reasons. First-In models assume higher value earlier in the chain because those touches are assumed to be “more motivating”.
Chain-based attribution doesn’t require assumptions about motivation, but instead supports highly fragile early auction-based touches because that is what is required to maintain chains that the data show create value.
Contribution Replaces Credit
With Chain-Based Attribution, contribution replaces credit in internal discussions about the deployment of resources. Marketers must recognize that combinations of channels, programs, and budgets create desired outcomes.
While this may sound daunting, the fact that Chain-Based Attribution puts outcomes first in valuing touchpoints, and only values the entire chain, actually makes it easier to have discussions on budgets, programs, and new initiatives that cross operational divides. Departments have to coordinate to create explicit chains of multi-touch engagement.
Chain-Based Attribution provides a clear outcome-driven philosophy of attacking attribution that can lead to readily apparent tactical approaches that maximize the value in multi-channel engagement.
Losing the fights over credit is just a bonus.
Opinions expressed in this article are those of the guest author and not necessarily Marketing Land. Staff authors are listed here.