Cross-Channel Attribution is Dead
Let me clarify something before my friends who work for cross channel attribution companies try to kill me. I don’t think that the market for cross-channel attribution is inherently dead, but I do believe that the way most vendors offer those services today is dead.
I believe that the markets for marketing mix modeling and cross channel attribution are going to merge into one.
The growth in cross channel attribution was spurred by the shift in advertising spending from traditional media to digital media. A few years ago, several startups saw a business opportunity to utilize the digital touch points generated by digital media to create models that could identify the drivers of sales. Then these startups decided to include offline sales channels and—voila, we got cross channel attribution.
Cross channel attribution is a bottom-up technique in which an algorithm analyzes individual touch points when our brand interacts with the customer—for example, when the customer clicks on a paid search ad on Google—and then allocates the credit of the purchase to each touch point proportionally to its impact in the purchase process.
In a sense this is similar to marketing mix modeling because the final goal is the same, which is to assign a return on investment on each marketing activity. But while marketing mix modeling takes a high-level, top-down approach, cross channel attribution takes a bottom-up approach.
The problem is that traditional cross channel attribution faces a problem of missing variables. For example, for broadcast marketing vehicles, like radio, we don’t have a way to individually identify who was in the car with their radio on at the time when we aired a commercial. Moreover, the models don’t normalize for changes in the economy, the weather, our competitors advertising spending, or other factors that affect our sales but are outside of our control.
As a consequence, the recommendations from most cross channel attribution projects are misleading, and therein lies the opportunity for a combination of marketing mix modeling and cross channel attribution.
I already see signs from the vendors that confirm this trend. Some companies that specialize in marketing mix modeling have purchased cross channel attribution solutions and some cross channel attribution vendors are offering marketing mix modeling. More importantly, these companies are working on integrating the results and methodologies of both techniques into a single solution.
The tipping point will occur when clients realize that there is no point in setting up a standalone cross channel attribution that doesn’t account for all drivers, including the ones that can’t be traced back to a particular individual.
Helping You Create a Marketing Machine for Your Business
6 年Damian, Good article. Actually, I think there are a few tiers of requirements, starting with Last Touch Attribution, somewhere in the middle is Cross-Channel Attribution, then there's MMM, then theirs MMM with brand health metrics.
Digital Marketing Strategy | Paid Media Optimization | Content Creation | Data Analysis
6 年Hi Damian - Great article! I completely understand your standpoint. I work for a company that, believe it or not, combines the best of attribution and MMM, resulting in a unified measurement model that measures and optimizes campaigns in-flight and gets down to the creative and person-level!
Chief Executive Officer at QuantaValue LLC
7 年Yeah, but that's hard. The old way is a lot easier!
Independent Consultant | Digital Strategy, Implementation
8 年Damian, great article.
SVP AI Enablement @ Blend| MIT Certified AI Strategist
8 年Great thread. I would also add to the mix that traditional linear statistical equations used to derive impact in a traditional MMM or MMO (top down) do not fit well in a digitally pervasive world in which everything impacts everything (more of a circular model) in which structural equations are a better fit and can be used to optimize more than a few goals (rev, brand love, engagement, social impact, etc...). Not to mention the use of single channel optimization techniques that produce lift and conversion in ways that may not be understood by a centralized model. Simply stated the math in middle of top down and bottom's up world also needs a major overhaul.