Advertising can somewhat cynically be defined as the process of convincing people to buy the next new thing. In advertising measurement, that new thing is MTA, the measurement technique touted to revolutionize how we look at analytics. It’s supposed to solve the frustrations we experience with other forms of marketing attribution: too costly, too historical, simply too slow in this age of moving-at-the-speed-of-light media interaction. While MTA adds significant value for given situations and businesses, it’s not a one size fits all solution. At the same time, Media Mix Modeling remains an incredibly effective and efficient tool for predicting and optimizing marketing spend. Both have their strengths. It’s a matter of knowing when to most appropriately use them.
The Perils of Using MTA Alone: Is Your Marketing Really ALL Digital?
For those who have a need for quick and targeted planning and have a sophisticated digital footprint and robust data warehouse, MTA may be their first choice for modelling.
When a client of ours in higher education wanted to conduct a purely MTA analysis for their business, we worked with their internal team to evaluate the approach. TV and paid social were key investment areas for them and an MTA approach would not capture those tactics. Their business is also sensitive to other non-addressable and external factors like changes in the economy, scholarships on offer, PR mentions and other factors MTA models cannot accommodate.
We suggested a unified approach, which provided a holistic sense of incrementality in the context of their business landscape. At the same time, we were able to deep-dive into more granular insights. This holistic measurement approach helped them materially grow their business. The model we developed identified up to a 5% growth in inquiries and applications and 3% growth in enrollment with the same spend. A potential increase in the marketing budget fuelled growth throughout the customer journey growing inquiries by 20%, applications by 10% and enrollments by 5%. The client got those incremental funds because we helped them demonstrate the precise value of marketing.
What does a unified measurement process entail? In this diagram, you can see where a comprehensive Media Mix Modeling approach (that incorporates person-level digital data) is augmented with granular qualitative TV insights (always measure the impact of the TV creative, not just the exposure) and the impact of both display and paid and natural search.
AP Solution: Unified Marketing Impact Analytics
A unified approach is critical to appropriate measurement. MTA in a silo may lead to biased results and poor business decisions. It could over and understate the impact of marketing if external or non-addressable factors that are not controlled for.
User-Level Data is Not Always Necessary to Deep Dive into Digital
Part of the excitement around MTA is its ability to go beyond broad strokes of mass media exposure to get to the individual user level. But this level of granularity may not be what is needed. The truth is, with a thorough and sophisticated approach to data and modelling, marketing mix studies can provide context into performance by digital channel, platform, site and campaign.
In recent discussions with a beverage brand, the marketing team believed they needed to leverage MTA and user-level data alongside their marketing mix models. In discussing further, we found that the questions they were looking to answer were focused around performance by type of digital (e.g. display vs. online video) with a deeper understanding of genre performance. For these purposes, MTA was not actually needed. They were better served by leveraging a digital deep dive as part of a marketing mix using aggregated data. We rescoped and refocused the engagement, which resulting in a targeted analysis that answered the questions raised.
Trying to determine MMM vs MTA, and when you are best served with both? The following table shows the sorts of questions you can answer through both types of attribution. Recognize what you are trying to accomplish?
Analytic Partners uses the following framework for describing which approach is most effective for various measurement goals.
MTA is best when:
- Used as part of a unified measurement engagement.
- When the business questions at hand necessitate user-level data to dig deeper.
- When there are other addressable channels outside of digital, such as email, direct mail or even household level TV data.
MMM is best when:
- Evaluating the performance of marketing channels in context of everything that may drive the business: ie weather, economic factors, in store activities, sales force actions.
- Aggregate data is sufficient to ensure deep dive measurement within marketing channels.
Unified is best when:
- The business has a sophisticated digital footprint, and an integrated planning process that starts with a more aggregated holistic view (the top down approach) and provides a more granular view (the bottom up approach).
- The business needs an understanding of incrementality of their investments and drivers of the business, as well as a greater understanding of the customer journey at a very low (placement, keyword) level.
- When you have a robust data warehouse that allows for linkage across users.
- Preeti Mehta, Senior Director, Client Engagement Team at Analytic Partners