February 26, 2021
John Wanamaker, a 19th century American Merchant and political figure once famously said: “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.” Marketing attribution was a problem even then.
Now fast forward to the 21st century, and we see that marketing is so much more sophisticated, and most marketing decisions are data-driven. Given the access to mountains of aggregate and customer-level data, it would be reasonable to assume that we have solved the problem of attribution. Not really, attribution is more difficult today than during Wanamaker’s era, which was dominated by a few marketing channels, namely print, out of home and radio.
Today brands are marketed across a vast number of channels and platforms, both analog (offline) and digital (online), generating data exposing almost every aspect of consumer behaviour. However, organizing this data and making sense of consumers increasingly non-linear paths to purchase is extremely complicated.
Despite the improvements in the “science” of marketing attribution, there continues to be confusion around the best way to quantify the impact of omnichannel marketing activities on financial outcomes. CMO’s struggle to provide an accurate demonstration of marketing ROI.
It would be great if marketers could drill down into insights from attribution tools in multiple ways. Some work with web analytics, third party analytics platforms; some develop proprietary attribution models, and some focus on a hand full of metrics. However, the holy grail of attribution would be the ability to accurately track the customer’s journey at all touchpoints through the entire marketing funnel from brand awareness down to conversions.
Three different approaches common today are A/B Testing, Market-Mix Models (MMM) and Multi-Touch Attribution (MTA).
Marketing Mix Modeling (MMM)
The IAB defines Marketing Mix Modeling (MMM) as a statistical analysis of aggregate sales, marketing, and business driver data that quantifies the impact of different marketing channels and tactics (the marketing mix) on financial outcomes over time. [i]
Marketing-mix models typically rely on market-level data and perform better with at least three years of marketing data (input) and sales data (output). Input data include measures of economic vitality and competitive activity, price promotions, distribution media channels and impressions and other on and offline tactics. Output data consists of the sales volume and value.
The resulting insights can be used to optimize marketing by reallocating spends to more productive channels and tactics and also predict future outcomes.
MMM attempts to answer critical questions like, what was the marketing ROI from a specific channel, and what would be the impact on sales if the spend was reallocated.
Multi-Touch Attribution (MTA)
In today’s omnichannel marketing era, customers interact with multiple touchpoints of the same brand on their path towards purchase. Attribution models let marketers decide how much credit each digital touchpoint gets for a conversion. Multi-touch Attribution models (MTA) provide a better understanding of how your ads perform and can help you optimize across conversion journeys.
The objective of the Multi-Touch Attribution model is to measure the impact of marketing activities on the metric associated with conversion and generate insights to guide decisions about future marketing spend allocation by tactic at an individual granular level.
The critical question MTA seeks to answer is, what is the expected incremental change in propensity to convert as the result of an impression of a specific touchpoint in the customer journey.
There are several attribution models, but not models are multi-touch. For example, first and last touch models are forms of single-touch attribution. Single-touch attribution models only factor in the first or last touchpoint encountered before a conversion, while multi-touch models evaluate the impact of several touchpoints.
Which model should you use, MMA or MTA?
Neither technique may be a complete solution in every case; there will be some blind spots. Both the advanced statistical methods provide value, their value may differ by the sector in which the brand operates and the channels the marketers deploy.
MMM works with historical market-level data and guides significant strategic decisions like spend allocation across the marketing mix. Whereas MTA primarily evaluates digital touchpoints, but at an individual level and in real-time.
For all the benefits, the MMM and MTA do no adequately measure the impact of the exogenous micro and macro environment factors, nor segmentation, brand and messaging strategy.
Omnichannel marketers with substantial investments across many channels and tactics may try the hybrid approach and develop a Unified Measurement Model (UMM) or Unified Marketing Measurement (UMM) as it is also known.
UMM integrates more than just MMM and MTA. It does this by assigning a business value to each strategic and tactical factor that influences marketing performance across all analogue and digital customer touchpoints, including messaging at the individual level. UMM is still new, and its complexity is better suited to the marketer with more significant resources.
How do you know your marketing is working for you?
[i] IAB. (2019). The Essential Guide to Marketing Mix Modeling and Multi-Touch Attribution. IAB. Retrieved from https://www.iab.com/insights/the-essential-guide-to-marketing-mix-modeling-and-multi-touch-attribution/