As Bob Dylan once said, “The times they are a-chang in.” Today’s marketing landscape is evolving at an unprecedented rate, and the pace is accelerating. This, in turn, has resulted in severe fragmentation and diversity of the channels through which people obtain information and content. Consumers move seamlessly between screens, platforms, channels, places, moments, mindsets and motivations— which is making the consumer’s journey dynamic and multi-dimensional.
However, as marketing evolves, though, measurement is becoming more challenging. For most marketers, improving marketing ROI is an evergreen objective. Increasingly, CMOs must demonstrate their contribution to the bottom line and make every dollar count. The ability to quantify ROI and demonstrate real business results is imperative, as is the need for continuous improvement in marketing performance. The goal is simple — understand the incremental business impact of marketing across all channels. Achieving this critical objective is anything but simple. The process to determine the return or incremental sales is often called attribution.
"Marketing attribution is a journey – look to make progress rather than aiming for perfection"
Marketing attribution is versatile and can be defined in many ways. Put simply, marketing attribution is the analytical science of determining which marketing tactics (audience, channels and messages) are contributing to sales or conversions, thus improving marketing ROI. Marketers today rely on a variety of attribution approaches – Marketing Mix Modeling (MMM), Digital Multi-touch Attribution (MTA) and/or TV Multi-touch Attribution (MTA) – to attribute revenue to, and measure the financial impact of, marketing. To date, marketers have used either to understand the effectiveness of marketing and make more informed investment decisions. All Each of these approaches has its own unique advantages and limitations.
Marketing Mix Modeling (MMM)
MMM is a top-down, macro-level approach that estimates the impact of marketing on incremental revenue. MMM calculates the total effect that every marketing channel (both digital and offline) and its key dimensions (such as creative, product, and geography) have on incremental sales and other performance metrics, while controlling for exogenous (non-marketing) factors like weather, competitive activity, seasonality, holidays, pricing fluctuations, and overall economic conditions – that impact business performance.
MMM breaks down business metrics (sales) to differentiate between contributions from marketing and promotional activities (incremental drivers) vs. other (base) drivers. These factors affecting marketing mix can be defined as:
• Base drivers: Base outcome is achieved without any marketing activities. It includes brand equity (long-term impact of marketing activities), distribution, weather, seasonality, competitive advertising and more.
• Incremental drivers: Business outcomes generated by marketing activities
MMM is critical for marketers who want to optimize ROI holistically across all business drivers – online and offline. The insights it produces helps marketing leaders plan both short and long term, determine how to best allocate their budgets, compare year-over-year metrics, and better understand non-direct performance measures such as brand equity/affinity. While MMM provides a picture of aggregate channel performance, it doesn’t provide the timeliness or level of detailed information that are required to support tactical decision-making.
Digital Multi-touch Attribution (MTA)
MTA is used for a bottom-up analysis of marketing investment. By leveraging individual, user-level data (actions and behaviors) across addressable channels (such as online display and paid search), MTA calculates and assigns credit for a conversion event to the marketing touchpoints that influenced a desired business outcome. Ideally, the data will include every exposure that an individual has had to a marketer’s messages and his or her responses (or lack thereof) to those messages.
There are many different approaches to attribution that range from basic, single-factor models (that rely on simplistic pre-set rules) to advanced models, which can incorporate advanced statistical techniques and complex algorithms. The level of sophistication of each can differ dramatically, and each method of attribution has its pros and cons.
But even the most robust algorithmic models vary in terms of their sophistication and granularity of predictions. Some algorithmic MTA solutions focus exclusively or primarily on estimating the financial impact of digital marketing activities, and they have limited ability to account for traditional media; external market forces; other critical marketing decisions like pricing, product, inventory, distribution & customer experience; brand building initiatives; baseline of revenue that would exist without any marketing efforts. Therefore, MTA solutions can overstate the amount of revenue attributable to digital marketing programs. Additionally, algorithmic MTA has been plagued by severe data reconciliation issues, walled garden blind spots, the collapse of third-party tracking, and challenges of identity and GDPR.
TV Attribution assess the immediate impact of TV advertising (across national, local, linear/live, VOD, OTT, connected TV, addressable) – by using causal inference measurement – and determines which elements of the TV buy (reach, frequency, network, program, genre, day part, day of week, DMA/geography, creative type or length, etc.) drive the greatest responses or actions (like search queries, website traffic/leads, website conversions/sales, or store visits/footfall) within seconds, minutes, hours or days of a TV ad airing.
TV Attribution applies a multi-dimensional, algorithmic approach to analyze response activity minute-by-minute and measures the additional levels of consumer response which was directly attributable to TV media investment over and above the ‘base’. The ‘base’ is the ongoing and underlying level of response which is the product of many factors, such as promotions, seasonality and the history of previous media investment.
Unified Marketing Attribution (UMA)
When it comes to marketing measurement, relying on any single attribution method simply won’t provide the insights needed to make the most informed decisions. MMM, digital MTA and TV attribution play different roles in planning, budgeting, executing, measuring and optimizing marketing initiatives. This is because marketers have different needs for information, at different speeds, based on where they sit on the team and the business challenges they are attempting to solve. While there isn’t a one-size-fits-all strategy for marketing measurement, attribution and optimization, any brand that relies on online and offline channels should consider all three approaches. Because of the distinct advantages of each approach, marketers realize the highest returns when they are used in tandem to understand the impact of marketing elements—media, creative and audience. Working with MMM, digital MTA and TV attribution side-by-side isn’t trivial, since they work on different levels of data, use different measurements for evaluating ROI and (typically) produce contrasting and conflicting insights.
Rather than settle for the individual models and their limited vantage points, and instead of forcing these three approaches to work together, progressive marketers should move toward a unified marketing attribution (UMA)framework methodology or what I like to call a Marketing Performance Management System (MPMS) – that leverages the strengths and eliminates the weaknesses of the three attribution approaches. It’s absolutely critical that an UMA is able to isolate the (1) short-term impact (promotional &campaign) from long-term impact (brand based), (2) the creative power (quality) and message strength effects from weight or investment and mix effects and (3) impact of brand equity on sales. It’s through combining these methods that marketers can achieve a more holistic view of the key drivers of marketing performance, and improve the ROI.
This ‘single source of truth’ model is more a complete, accurate and actionable understanding of what is driving the business at the strategic and tactical levels (rather than over- or under-attributing the impact of marketing), guiding high-level allocation and long-term planning decisions, as well as near-term, discrete (tactic-level) granular optimization choices by tactic, creative and audience.
At a time when marketers are increasingly held accountable for results, any attribution model is better than no measurement at all. While there is no absolute right answer for all marketers on which attribution solutions to pursue and in what order, it is incredibly important that marketers craft an integrated attribution solution to fit their business. Ultimately, deciding which attribution solution is right depends on goals, business requirements, how the output will be used to improve marketing effectiveness, organizational readiness and data availability and quality.
Marketing attribution can be time and resource intensive to get right. However, when done effectively, attribution brings a myriad of benefits that fall under the following key drivers of unlocking demand potential, growth(both brand &business), and economic value: improving marketing effectiveness, increasing marketing efficiencies, and maximizing marketing ROI (financial performance). The end-result is increased accuracy, clarity and confidence in decision-making and greater marketing accountability. Such decision tools do more than provide marketers with valuable information. They stimulate dialogue about real (short- versus long-term) trade-offs, drive thoughtful ‘what-if’ scenario planning and help to manage expectations across business units and functions whose cooperation is often critical when companies change the broader commercial mix.
Marketing attribution is a journey – look to make progress rather than aiming for perfection. Marketers need to realize that “perfection is the enemy of progress,” to quote Winston Churchill. A perfect state rarely exists – and marketers should not wait for perfection before making optimization decisions. Marketers need to understand that there is no exact predictive science on human behavior. Werner Heisenberg’s (an early 20th-century theoretical physicist and pioneer of quantum mechanics) Uncertainty Principle is all about trade-offs in our ability to take precise measurements and generate insights on consumer behavior and to know when to be cautious about trying to predict that behavior.