About six months ago, I jumped from an exciting agency-side career to join the Marketing Science team here at LiftLab. It’s a shift I’ve benefitted from for several reasons – not the least of which because of how much I’ve learned. To be clear, I encountered incredible levels of knowledge and sophistication in the agency world every day, from peers to mentors to the methodologies we used. But there’s an inevitable leap in subject matter expertise you’ll get by going vendor-side – especially if it’s a vendor with some of the leading minds in marketing measurement here to guide the science.
With those learnings in mind and agency practices still fresh in my memory, I wanted to share some of the learnings that particularly struck me that even the most sophisticated agencies could benefit from putting into practice.
I plan to write about many of these in an occasional series. To start, I’ll focus on the relationship between incrementality and mix modeling.
Mix Modeling and Incrementality: One Practice Area or Two?
In my experience, many agencies manage two major analytics functions concurrently: an incrementality testing practice – managed by either an in-house team or a provider; and, separately, a Marketing Mix Modeling (MMM) initiative, typically managed externally. On the surface, the logic is solid: these are two different arenas – one taking snapshots of the incremental impact of marketing changes, and the other looking at a 10,000-foot view of that data and how it impacts a business over time. It’s only natural that you’d expect to deal with each expertise distinctly, especially given the varying levels of lift involved in doing so. The surface logic is undeniable. But after working at LiftLab – which executes both MMM and experimentation on one system – I’ve come to realize just how much the two practices are inherently linked, and how much unnecessary friction agencies create by treating them as distinct entities.
Experiment design is inevitably guided by marketing mix models; and, in the best MMM approaches, experimental findings are used to make those models more intelligent. Keeping these practices distinct means agency teams are consumed by constant work liaising between teams and providers, porting data across them, and losing valuable time and insights amidst all the coordination. By connecting experimentation and MMM within teams and software, agencies can create a cohesive measurement lens that is self-teaching and efficient—leading to more effective experiments and models and faster time to knowledge, while also freeing up resources to focus on brand-specific insights like audience overlap and customer lifetime value.
A Unified Measurement Approach for Agencies
Connecting experiments and MMM is foundational to our approach at LiftLab, and it’s a linked approach that I think agencies could benefit from immensely. One analysis finds that “71% of agency professionals believe their job is harder today than it was just two years ago” and that “inefficient processes…and siloed/disconnected systems are among the top challenges currently facing agencies”, while another study finds that, more generally, 95% of senior IT executives at mid-sized to large companies are looking to consolidate their tech stacks, primarily to reduce the number of point solutions and work with a handful of “strategic vendors.” That external research matches my own experience: Agencies are the often-unsung heroes that contribute incredible value for the whole marketing ecosystem by supporting a dizzying number of functions every day. Having experienced a different way of achieving one of those core functions – marketing measurement – I thought it was just too efficient not to share.
