For modern marketers, the allocation of scarce marketing resources to help maximize business performance is fundamental to success. Consequently, attribution and the role of marketing investments in driving consumer demand are at the top of the agenda.
Never has this been more apparent than in today’s multi-channel marketing environment, where the seemingly endless proliferation of both off- and online channels has significantly complicated the measurement process. Attribution methods have, therefore, had to evolve accordingly.
The dominance of MTA and MMM
Out of many possible approaches, two in particular have come to dominate the field of marketing attribution: multi-touch attribution (MTA) and the marketing mix model (MMM).
MTA is a “bottom up” approach, focusing on the contribution of online touchpoints to an online conversion outcome across a network of cookie-level pathways.
MMM, on the other hand, is a “top-down” approach, focusing on nested systems of econometric equations with pricing, paid, owned and earned media working together to drive demand.
Modern challenges in marketing attribution
The role of these two techniques in the next generation of marketing attribution is something our group explores in a dedicated chapter of the recently published 2023 I-COM Global Data Science Journal. Here we set out how well-documented problems surrounding user identifiers, walled gardens, GDPR and third-party cookies have seen customer-level MTA fall out of favor in recent times.
The Renaissance of MMM
This has led to a resurgence in the popularity of MMM, notably among key players such as Google and Facebook, where aggregating across consumers can help resolve privacy issues and capture a far wider set of demand drivers such as price, offline marketing, and economic factors. However, to constitute a valid attribution framework, any viable MMM needs to address three fundamental criteria: causal inference, short and long-term measurement, and granular “real-time” insights.
The causal dilemma in attribution
Firstly, marketing attribution and budget allocation rely on accurate causal attribution to each element of the marketing mix. Modern MMM attempts to inform this process via path-to-purchase theories of demand, where paid, owned and earned media work together to drive sales.
However, much like MTA, this approach ignores the fundamental problem of selection bias in online media such as paid search. This occurs when (part of) the sales outcome is caused by a factor that predicts the likelihood of selection into paid search rather than due to search itself.
This creates an identification problem, leading to biased estimates of the search-sales impact and all marketing effects that work through it. Many solutions have been proposed, ranging from purely statistical techniques such as instrumental variables, Difference in Differences and Gaussian copulas, through to the use of experimental results as Bayesian priors. Whichever route is taken, the message is clear: to isolate true incrementality, the chosen identification scheme needs to be clearly specified as part of any MMM engagement.
Balancing short and long-term impacts
Secondly, for a complete view of marketing ROI and optimal allocation, marketing mix models need to reflect both short- and long-term marketing effects. Short-term effects explain temporary or transitory sales variation. Long-term effects explain persistent changes in underlying base sales reflecting permanent additions to the loyal customer base.
Measuring the true long-run impact of marketing investments, therefore, requires a mix model capable of quantifying how marketing drives long-term base evolution, such as that set out in our latest IJRM article. Standard mix models with fixed baselines are not set up to do this and just amount to ad hoc extensions of the short-term model structure – typically comprising either Adstocks with very high retention rates, the addition of attitudinal brand metrics or simply multiplying the short-term effects by a chosen scaling factor.
Addressing granularity in modern attribution
Finally, next-generation MMM needs to fill the gap left in a cookie-less world to deliver granular and swift insights on marketing ROI and optimal budget allocation. Suitably identified high-dimension mix models – across consumer cohorts by day or hour – can fit the bill.
This can provide many of the claimed benefits of MTA, such as granular online media effectiveness ranking by publisher and placement, together with the ability to measure the impact of pricing, offline media, economic factors, and longer-term brand-building.
Navigating the future of marketing attribution
As the landscape continues to develop and we come to appreciate the off-online customer journey on ever deeper levels, marketers need a trustworthy attribution solution. Unsurprisingly, in the wake of the MTA revolution our collective standards have become far more exacting in response to increased complexity and depth. Yet, we do ourselves no favors if we let these issues inconvenience us or stand in the way of meaningful progress.
Embracing a sustainable path forward
So, we make a practical and highly strategic choice. Since the many challenges facing MTA are unlikely to be resolved, the marketing mix model is apt to be the most sustainable approach going forward. Provided the ongoing development and practice is rooted in firm academic foundations and flexibly adapts to commercial needs, modern MMM can help drive productive and thriving, long-term sustainable business performance.
Attribution is one of the most important aspects of your success as a CMO. Join the conversation around it on the CMO Alliance Community Slack channel and connect with a global network of marketing leaders and CMOs.