When traditional “database marketing” first took off in the early 1990’s, marketing performance measurement and attribution was quite simple. We generated sales and direct mail campaign performance reports using a handful of dimensions. Attribution was easily derived through business reply cards (attached to direct mail pieces), phone numbers or tracking codes. We also used indirect attribution rules by making control group comparisons. We were fairly accurate and the process was easy to execute.
The Current State of Attribution
We all know that the marketing landscape has changed … and it continues to evolve with massive channel proliferation. With so much data and so many options regarding how to best apply a limited marketing budget, how can a CMO receive richer insight to influence tactical decisions that will improve media/channel performance?
Let’s first examine the various states of attribution from the viewpoint of the modern day marketer:
- Direct Attribution: Still used widely today and still relevant. A specific customer behavior (e.g. a purchase) can be “directly” attributed to a given marketing stimuli via a unique code, landing page/URL, response device, etc. However, other marketing stimuli may have created momentum and been a significant contributor to the consumer’s ultimate decision to purchase.
- Last Touch Attribution: Attributing the desired customer behavior to the last “known” marketing touch. Similar to “Direct” Attribution, but not always the same, here the marketer attributes the desired customer behavior to the last known touch. This method is very common when there are no specific tracking codes/tags that tie a desired customer behavior directly to a specific marketing stimuli.
- Multi-Full Attribution: Channel proliferation has led to individual channel/media silos, each with their own unique attribution rules. The separation of traditional offline data and online data is very common. For example, direct mail data is stored in a traditional customer database, email data is stored with the email service provider, and online data is stored by various DMPs, by vendors/partners that are contracted to capture it, each often with their own siloed attribution logic taking FULL credit for the same desired behaviors.
- Rules Based Attribution: Building on the “Multi-Full Attribution” described above, here marketers use what is often called a “common sense approach” to proportionally assign attribution to very siloed marketing stimuli. For example, a business had recently identified the large overlap between their direct mail and digital channels. For the overlapping purchases identified in both groups, 100% of a given purchase was attributed to direct mail, while simultaneously 100% was also attributed to a combination of digital channels. A rule was then quickly implemented to assign 20% of the attribution to the direct mail channel and proportionally reduce the attribution by 20% across the various forms of digital media. So, it is “fractional” by the simplest definition, but no real math or analytics was being used to assign the “fraction” to each media/channel.
Each of these options contains significant attribution bias towards channels/forms of media, that when taken for face value will result is less than optimal decision-making.
What’s Next and What is Fractional Attribution?
Marketers must now leverage math, science and statistics to analyze and derive insight from large pools of data, much of which can now be integrated across channels to inform decisions across touch points during the customer journey. Fractional Attribution is a necessary tool for understanding campaign performance across a multitude of touch points.
Through advanced (and proven) analytic techniques, a weighting calculation is developed and applied to the various marketing touches during the customer’s buying journey. In short, you are attributing a portion of that customer’s purchase to each of the marketing touches that impacted the customer’s decision to buy.
Harte Hanks has a team of analysts that work with marketing organizations to create a fractional attribution model through a collaborative development process:
- Define the overall objectives and identify the behavior metrics you want to positively impact (e.g. response, sales, conversion, product registration, etc.).
- Define and implement the roadmap including identification of key performance indicators (KPIs) and setting the overall attribution approach. Companies have used both “quick start” fractional attribution solutions and more robust solutions that require dedicated data stores and data integration tools.
- Collect and compile the data.
- Execute the fractional attribution solution and create the scenario planning tool.
The “scenario planning tool” is what enables the user to optimize media/channel performance. Using the tool, the analyst or marketer can quickly run “what-if” analyses to estimate the impact of reallocating marketing spend across channel/media or removing a channel/media from the mix altogether. The end result is a much more informed decision that can result in significantly higher returns from your marketing budget. Performance data and insights from the optimization exercise are then used to calibrate and refine the attribution engine going forward.
Fractional Attribution rooted in proven math and statistical techniques is a critical tool to accurately improve and optimize the performance of an incredibly fragmented and complex system of channels and media, both online and offline.
It’s not perfect – no marketing science or advanced marketing analytic solution is. But a robust modeled attribution solution is proven marketing science, and those that leverage it appropriately will generate higher return from their marketing spend and outperform their competitors.
Has your company used fractional attribution to better analyze your marketing spend? Share your experience with us in the comments section.