Wayne St. Amand is CMO at Visual IQ, a provider of marketing intelligence. He presented “Multi-Touch Attribution Across Facebook, Digital and Beyond” during the RampUp Virtual Summit on People-Based Measurement.
It has been over a decade since marketing attribution capabilities first came onto the market to measure the performance of marketing and advertising.
The objective of attribution itself is relatively simple: to accurately assign credit to the channels and tactics along the consumer journey that contributed to a desired outcome. Marketers can then make better decisions based on a fuller understanding of what’s working, and what’s not.
Yet many marketers are still confused as to how attribution fits into their overall marketing strategy. With a number of options available, it can be difficult to determine which measurement approach is best for your business.
So, where should you begin on your marketing attribution adventure? To answer this question, it’s important to first understand that marketing and advertising tactics have different levels of granularity, and that different attribution approaches have varying degrees of sophistication.
For example, understanding yesterday’s top performing display ad size for your business using in-store sales as your key performance indicator (KPI) is very different from attempting to understand the impact of each of your channels on a quarterly basis using total sales as your KPI.
For these reasons, multiple attribution approaches exist. Let’s explore:
Marketing Mix Modeling (MMM): MMM involves the collection of large amounts of summary-level historical performance data in an effort to reallocate budgets across channels and subchannels using various forms of logistical regression modelling. This approach also allows for the modeling of exogenous data, which helps marketers understand the impact of factors that fall outside of their control, such as the weather, economic conditions, and competitive strategies.
The benefit to MMM is that it can be used across both online and offline channels to incorporate a brand’s entire marketing mix. The downside is the level of granularity, speed, and consulting involved. Modeling occurs at only the channel/subchannel level on an infrequent basis (typically quarterly), and requires the involvement of data scientists, consultants, and other service professionals.
This dramatically impacts a marketer’s ability to optimize their marketing and media, as all recommendations occur at a summary level, with no granular understanding of tactical marketing performance.
TV Attribution: TV Attribution starts by establishing a baseline for digital channel performance using multi-touch attribution (I’ll get to that) without TV present in the marketplace, and then quantifies the incremental impact of TV impressions on digital responses using logistical regression modelling.
Most vendors can provide fractionally attributed metrics by TV dimension, including network, program, spot length, creative, geography, etc. Since most advertisers view TV as a top-of-funnel activity heavily focused on brand-related marketing, TV attribution is an effective way to understand how TV drives digital brand engagement key performance indicators (KPIs) like search queries and site visits, as well as final conversions. The modeling cadence for TV attribution can be relatively quick (weekly) and once implemented, uses an automated software-as-a-service (SaaS) technology to eliminate manual modelling practices.
Multi-Touch Attribution (MTA): MTA focuses on addressable channels and solves the granularity problem that MMM cannot address.
A unique identifier (UID) is used to track users across channels to understand the effectiveness of marketing tactics at extremely granular levels, such as keyword and placement. This happens at a much faster modeling cadence, with a full rebuild of the model occurring daily. MTA allows for multiple types of online and offline KPIs to be established by syncing UIDs. Data is then automatically ingested and modeled using SaaS technology with optional levels of ongoing support post-implementation. Fractionally attributed data can be integrated directly into programmatic buying platforms to allow for near real-time optimization, and UIDs can be synced with a host of different first- and third-party data sources.
The challenge with MTA is that not all marketing tactics are addressable, which means that fractional credit is only allocated to the marketing tactics it can track. It must be integrated with MMM for a truly comprehensive view of performance.
The Bottom Line
While MMM, TV Attribution, and MTA are different marketing attribution approaches that meet different needs, all will arm you with insight into performance that you didn’t have before. Depending on your organization, you may benefit from one, two, or all three approaches. Ultimately, deciding which approach(es) are best comes down to your goals, business requirements, and how you want to use the output to improve the effectiveness of your marketing.
For more industry insights, listen to Wayne’s session from the RampUp Virtual Summit: