How To Measure And Attribute Marketing Performance Across Channels

Every marketer wants to understand the effectiveness of their efforts. But historically, getting reliable information and insights has been difficult. This post outlines the challenges and solutions surrounding cross-channel attribution and moving beyond last-click.

The Dark Ages

Before digital, it wasn’t easy to attribute business performance to the tactics that drove it. With old-school, non-addressable television, we either tried to correlate activity with campaign flights or used (sometimes laughable) artificial research methodologies to determine whether our ads were working. The challenges were even more significant with Print, Radio, and Outdoor. We had little to go on until research companies began to offer “Media Mix Modeling” – a $ix-or-$even figure solution that took months to complete.

Enter Last-Click

When digital entered the mix, people were thrilled that we could directly tie actions back to tactics. Most companies adopted “last-click attribution.” For the first time, we could identify relationships between investment and results rather than guess or infer.

The wonderful thing about last-click was the direct connection between a marketing initiative and a desirable event. It is also easy to calculate. But it has many shortcomings, not least devaluing every marketing event before the final touch. Based on this methodology, for example, paid search is “over credited” because people often use it to find a place to buy things (“navigational search”) after they have completed most steps in their journeys. One of the reasons why 41% or $78B of total digital spend goes to paid search is because of continued reliance on last-click.

Many in the affiliate industry loved last-click because our leading publishers are optimized to capture that last touch. But there’s now widespread recognition that last-click does a disservice to many classes of partners that deliver great value to brands. Further, some suspect that certain affiliate vendors began aligning their tactics to getting credit through last-click rather than maximizing genuine business value. For example, much of the controversy surrounding extensions and toolbars is driven by the sense that some of them steal credit for conversions from the actual referring publishers.

Getting credit for a transaction is different from being the driver of it. So long as we cling to a methodology that EVERYONE knows is broken, we will continue to have managers and execs questioning the value of the channel. Including partner data in attribution models is critical to demonstrating that sales through this channel are profitable and incremental.

Multi-Touch Attribution

More recently, companies have focused on refining their attribution efforts, allocating credit to multiple touches in the customer journey. Most companies use SaaS platforms or service providers to “do” multi-touch attribution. Methodologies vary.

So far, relatively few companies have chosen to include the affiliate channel in their multi-channel attribution models. Why? First, we’ve been so wed to last-click. Second, until recently, many affiliate networks did not provide the data granularity necessary to incorporate the channel into models. But that has now changed.

For affiliate to become a core element of every brand’s marketing program, we need to integrate our channel into the company’s broader attribution modeling. Let’s look at how multi-touch attribution is calculated.

Rules-Based Versus Algorithmic Attribution

Multi-touch attribution solution providers – whether standalone solutions or components of an affiliate platform – have methodologies that fall into one of two camps:

• Rules-Based Attribution assigns a set amount of credit for an outcome to each touch. For example, the model might say each touch is equally important. Or that the first touch deserves ⅓ of the credit, the last ⅓ of the credit, and all the touches in between deserve a share of the remaining ⅓. The set fractions are “rules” that are applied to every transaction. Using set rules makes attribution straightforward. The rules are based on judgment, which may (or may not) offer accurate approximations of the causal value for each touch.
• Algorithmic Attribution examines each transaction and assigns credit based on data-driven analysis. It mathematically asks, “Did this event cause or contribute to a conversion, and how much? With algorithmic attribution, the allocation is not determined by set rules but rather by extensive data analysis. This methodology is significantly more complex than rules-based approaches but can be more accurate.

Since rules-based attribution is based on judgment, its value in delivering an accurate picture of causality is based on the quality of the judgment. Rules-based methodologies for allocating credit can also be “gamed,” so companies that use them need to be vigilant.

Cross-Channel versus Single Channel Attribution

If you and your company have decided to move away from last-click, you must choose a new approach. In partner marketing, some people think of attribution as a single-channel analytics challenge. But single-channel attribution isn’t a long-term solution because consumers don’t just interact with one form of media. Instead, they seamlessly migrate from touchpoint to touchpoint across media and devices.

In multi-touch, multi-channel customer journeys, partner marketing touches are intermixed with touches from other marketing programs.

Accurate attribution is all about understanding the interplay across media and devices. To gain this understanding, you must include a comprehensive data set on consumer events for the partner marketing channel and all other channels. Through advanced analysis, that data — when combined with performance insights across the marketing mix — helps to identify the relative importance of different touch points toward driving a conversion.

Some companies may not be ready to deploy multi-channel attribution. In those cases, you may consider single-channel attribution. Many in the industry say this is a more accurate approach than last-click.

What Affiliate Leaders Should Do

Here are some actions you can take to ensure that every affiliate touch gets the attribution it deserves.

1. Find out if your company has multi-touch, multichannel attribution in place. Work with the larger organization to incorporate affiliate data into the modeling if it does. You’ll need to understand how the system ingests data and the data requirements. Then work with your platform or network to get the data sets or APIs necessary to participate. If your company does not have such a model, it makes sense to begin conversations about it.
2. If multichannel attribution is not realistic for your organization, consider whether it makes sense to implement channel-specific multi-touch attribution. Talk to your platform or network about their single-channel attribution offerings. Most offer some form of multi-touch attribution modeling, though the quality can vary by provider.
3. If you pursue single-channel attribution, develop a strategy and test approaches before broad-scale implementation.
   • How urgent is moving away from last-click for your program? If you work exclusively with bottom-of-funnel traditional affiliates, the urgency is lower than if you are trying to work with other kinds of affiliates, especially those that drive greater brand awareness and consideration, like content sites.
   • Implement testing to determine how the new approach changes allocations and commissions. Do the changes make sense?
   • Find out if your top partners accept multi-touch. Some large ones won’t because their business models are built on last-click. If they don’t accept partial payments, you will need an approach that enables you to treat conversions differently based on the partner that drives the final touch.
4. Consider incrementality as you approach the challenge. Is incrementality important to your goals? Some publishers may drive lots of sales that you would have made anyway. Others may drive lower total volume but mostly net new buyers. It may be even more critical to adjust your commission rates by publisher or publisher class incrementality than your attribution methodology.
5. Keep your partners informed. Don’t make rash changes to your methodology without warning. Explain what you are trying to do and solicit ideas and opinions. If you decide to make changes, meet regularly to discuss how the new rules impact their earnings. Collaborate to find ways to help them maximize performance under your new model.

Getting attribution right is an important goal for any affiliate leader. You may have to take multiple steps to achieve true cross-device, cross-channel attribution. But any action that makes your allocation more accurate is a valuable step forward.