The Myths of Multi-Touch Attribution
With the inception of digital advertising and online conversion tracking, marketers have been mesmerized by the marketing attribution dream of “right ad / right time / right place / right customer”. This dream has some valid roots: targeting and personalization can make a meaningful difference by increasing relevancy and reducing inefficient ad waste. But, the marketing dream has launched a problematic perception that a brand must track every consumer across all ads and devices to know the value of a marketing campaign. Whether you call it multi-touch attribution, MTA, or some other tracking-based marketing measurement, this misperception is a core foundation to the many myths in marketing attribution, and OptiMine addresses each of these below.
Why does this topic matter? OptiMine still encounters misperceptions in the market that espouse or promote many fallacies about marketing attribution, and these are damaging. Marketers spend inordinate sums of money, create unmanageably complex systems, and waste precious resources chasing this measurement holy grail.
There’s good reason multi-touch attribution is in the trough of disillusionment. Gartner’s Hype Cycle1 for Digital Marketing2 shows MTA squarely in the trough. Expectations around the ability to map ad paths with confidence, control ad sequences across channels, identify and track consumer ad impressions and match identities across devices were way overblown. And when marketers attempted to deploy multi-touch attribution solutions, they found them extremely difficult to implement, costly to maintain, and incredibly brittle and inflexible. First adopters stung by these failures continued to search for other solutions in the mistaken belief that the issues were specific to the vendor selected, and not endemic to the methodology itself.
So, let’s explore these myths.
Marketing Attribution Myth #1: The Magical Path to Purchase is Key to Marketing Glory
It is very tempting for marketers to believe that advertising can be reduced down to an equation of ad exposures, and that by unlocking this secret formula, marketing glory and career success will follow.
AD1 + AD2 + ADn = Massive Conversion Lift ⮕ Glory, Promotion!
Marketers have expended vast efforts to make the dream a reality. Just consider some of the basic building blocks that are required to pull off multi-touch attribution to see why so many MTA efforts have failed:
Complexity Dooms MTA:
Also, there’s a common misperception that an MTA solution is “done” once the tracking code is in place. In fact, the solution isn’t even close to being done. Data must be collected over time to produce the first measures. Do you have major seasonality in your business? This data must be collected over time. Major promotions? This data must be collected over time. So on, and so forth. Oops! Need to make a change or collect new information? Go back to Start and begin all over again…
Marketing Attribution Myth #2: The Secret Sequence of Ads Must be Unlocked for Marketing Glory
When using a single marketing channel with high addressability (think email or direct mail), it is easy for a marketer to look at a sequence of “touches” and make some conclusions about an ideal sequence, contact pattern or cadence that leads to higher conversion rates. Stream tests that alter or adjust the pattern and timing of communications can compare one sequence with another to determine which yields the highest performance. This works in isolation within a single, addressable channel. But the reality is that most brands use 30 or more marketing channels and partners across a diverse marketing ecosystem.
And, why do we believe the sequence matters? If we look at our own behavior, do we see an ad on the fourth, fifth or sixth time and say to ourselves, “it’s time to buy!”. Of course not. Consumer behavior isn’t linear. And our needs, wants and problems that we need to solve- and thereby the products we buy- aren’t linear either.
Marketing Attribution Myth #3: Marketers can control the sequence of ads
Let’s pretend for a moment, that the sequence of ads across channels and devices mattered. And let’s also pretend that we could determine which ad was seen by a customer with high reliability (see Myth #4 below). Finally, let’s pretend that by knowing all of this, that we can assign an accurate value for each ad’s incremental contribution to sales.
Example: we know that a video ad on a mobile device that is 3rd in a sequence of 8 ads across TV, search, social and video contributes 17% to conversions. Great!
So, what is a brand going to do with that knowledge?
Will the brand somehow convince the consumer to see the video ad 3rd in an 8 ad sequence? Can a brand ensure that the customer sees their video relative to their other channels’ campaigns- at a consumer AND device level? Will the brand somehow convince Facebook to wait for their ad to be served after Google gets to go first? Come on.
This myth- that brands can control the sequence of ads- is really the most damaging misperception among the many with multi-touch attribution.
Not only is it not possible, but pursuing this myth creates a whole host of complexities and costs— all for what exactly?
Marketing Attribution Myth #4: Marketers can know every ad seen, across all devices for all customers. Bonus Myth: Consumers Like to be Tracked
Even with digital advertising, it isn’t possible for a brand to know every ad that a consumer has seen across all of their devices. Traditional advertising in TV, radio, print and OOH have always had this struggle, but with the advent of digital tracking, marketers have invested huge amounts in technologies and 3rd party data sources to pursue this tracking dream. For most brands that have a complex mix of digital and traditional advertising, and many different conversion touchpoints, data match rates are a significant hurdle.
In the simplest case where a brand spends all of their advertising budget in Facebook and Google, match rates will be better especially if the brand shares its customer database with the walled gardens. Why a brand would entrust their entire customer file with one of these players is an entirely separate (but excellent) question. But most brands execute their marketing across a much broader array of partners, channels, publishers and networks. And no single marketing campaign works in a vacuum. Cross-channel effects from one campaign and channel impact campaigns in other channels. Given these cross-channel effects, gaps in the ability map ads, consumers and devices create a major degradation of marketing measurement.
Taking a step back to look at the bigger picture, there is a critical role for the collection of personal information, and brands that invest aggressively in building out their first party data collections will perform better over time. There are many other priorities to help build the business case for PII:
- Personalization and personalized communications
- Message and offer targeting
- Loyalty programs and memberships
- Improved service and support
- Consumer experience mapping
But marketing performance measurement shouldn’t be the driver of a first party data strategy. Why? Because:
- It isn’t possible to know all ad exposures across a brand’s customer or prospective customer base
- The challenges and costs in pursuing ad tracking are getting worse (a LOT worse), not better
- Consumer device/ad match rates have always been poor, and with Apple’s privacy changes, they are suddenly way worse. And in 2022 with Google’s pending FLoC privacy changes, things are going to get very, very dark.
- Even small ad-consumer-device matching errors create huge model accuracy issues
Bonus Myth: consumers like to be tracked. It is clear that the US is entering a major new phase of consumer data privacy. Consumers are tired of their data getting hacked, are alarmed at the lack of transparency in the handling of their data and are increasingly taking steps to protect their privacy. And the consumer backlash is now driving major new consumer data privacy regulations, and therefore new risks for brands. With state-by-state consumer privacy regulations exploding, brands must now contend with privacy regulation risks precisely because of this breathless chase for tracking.
“Exploding consumer data privacy regulations are the karmic closed-loop of consumer tracking.”
Marketing Attribution Myth #5: Marketing Attribution Must Track Consumers to be Effective
It is time to re-define marketing attribution. Early on, the market and the analyst community was too prescriptive in terms of how the marketing attribution was to be accomplished, as opposed to focusing on the marketing objectives to be achieved by marketing attribution. This “marketing attribution = tracking consumers across their devices and ads” has done a real disservice to marketers because it has led them down a path of frustration, wasted time and effort and significant costs.
While focusing on how attribution must be achieved, the most important reasons to do marketing attribution have been lost:
- Measure the incremental contribution of marketing
- Deliver tactical, rapid reads of performance at campaign levels
- Provide guidance to the marketer on short term performance lift and optimization opportunities
Nowhere in the above objectives does it say how attribution is to be accomplished, and why should it? If a better methodology exists to meet these objectives and it doesn’t suffer from all of the pitfalls and PII risks, it is still attribution, and a better way at that. Marketers have been sold down a very specific path of attribution, and it is time to take a better path.
Where should marketers go from here?
Moving forward, path and tracking-based attribution will come under even more pressure as Apple has dramatically reduced individual tracking across its devices, and in 2022 as Google stops all individually targeted ads in Chrome. On top of this, a major wave of consumer data privacy regulation is expanding across the US beyond the initial California CCPA and CPRA regulations.
Tracking Isn’t the Only Way to Measure Marketing (and It Certainly Isn’t the Best Way)
Because of these issues, and many more, there are other ways to measure marketing performance that are far more accurate— like Marketing Mix Modeling— or MMM for short. Marketing mix modeling has been a proven part of large brand’s measurement toolkit for decades and has been used successfully to measure mass media effects on outcomes, while also accounting for non-media factors (seasonality, weather, economy, promotional pricing, etc.). The problem with traditional MMM is that it has been a manual, slow, expensive, and not very detailed set of guidance- not a surprise given that it is a technique from the low-fidelity days of the 60’s and 70’s. But modern, highly agile marketing mix modeling solutions (like OptiMine’s) built with today’s high-scale computing and AI are able to provide marketers highly accurate measures, in a fraction of the time, at far lower cost, and delivering high-fidelity detailed guidance.
Marketers must now get serious about future-proofing marketing measurement (see OptiMine’s future-proofing guide, here) as individual consumer tracking data becomes more and more incomplete. The disruption in the multi-touch attribution sphere will continue forcing brands to abandon this first generation of tracking solutions.
There is a better way: OptiMine. 100% future-proof. No PII ever.
1Gartner, Definition of Hype Cycle, 2021
2Gartner, Hype Cycle for Digital Marketing, 2020, Mike McGuire, Colin Reed, 15 July 2020