Marc Guldimann, CEO of Adelaide argues that letting media sellers optimize to outcomes at the expense of media quality, or worse, transacting on outcomes, creates a set of incentives and measures that sellers with sufficient data can easily game.
For decades audiences have been at the forefront of media value, something that is about to change as more audiences become addressable and the nuances of reach become measurable. The dialog now revolves around the first real data-driven discussion of media quality and its role in producing tangible outcomes for the marketer.
Lumen and Facebook have argued about the inclusion of attention metrics in those quality measures. Late last year, Lumen presented the attentive CPM, or aCPM, meant to represent the average cost of a thousand seconds of attention per platform. To compute aCPM, Lumen used eye-tracking to detect the average duration of gaze per impression and then factored in price. Facebook countered that duration of attention wasn’t worth measuring, calling Lumen’s tool “blunt” and suggesting it will “push the industry backwards”.
Before we dig into the arguments and offer a better solution, I’ll share a high-level theory of how we can use attention to measure advertising. Media placements, the containers that present creative to an audience, create the opportunity for attention and contribute to attention being held by limiting distractions. Then creative captures attention and holds it while presenting brand assets and hopefully nudging behavior.
At Adelaide, we call this theory the Attention Pathway. In this context, quality of media can be defined as the probability of attention to creative, discounted by likelihood of interruption. And the quality of creative is measured by its ability to hold attention long enough to introduce brand assets.
The Attention Pathway
Lumen’s approach, based on millions of seconds of observations in labs, joins a growing body of research supporting the idea that attention is the precursor to advertising impact. But attention, as measured by gaze duration, is a complicated affair that extends past media quality – nuance is required to disentangle the effects of attention on media and creative.
Duration also suffers from the inconsistent value of seconds. The first and fifth seconds of attention to an impression have very different value. Most research shows incremental yield dropping after two or three seconds of attention.
Finally, and to Facebook’s point, the per-second impact of different media is wildly variant. Karen Nelson-Field’s research shows that larger screens not only drive more impact than smaller screens, but that impact decays more slowly. For attention models to scale cross-format, they will need to consider the quality of duration and integrate a feedback loop with learnings from outcomes.
The aCPM is an admirable and data-driven start, far better than viewability or even duration weighted viewability at measuring media quality. Unfortunately, rather than iterate on aCPM or innovate in the measurement of media quality, Facebook suggests advertisers “measure outcomes, not exposure.”
Without question, driving attitudinal and behavioral changes (aka outcomes) is the reason marketing exists. All marketers investing in media should understand the relationship between spend and results. But this doesn’t need to mean relinquishing data. In most cases, media sellers shouldn’t be entrusted with attribution of outcomes – especially those with a track record of challenges around measurement and data integrity.
Letting sellers optimize to outcomes at the expense of media quality, or worse, transacting on outcomes, creates a set of incentives and measures that sellers with sufficient data can easily game.
Suppose an advertiser relinquishes targeting control to a seller with superior data, and agrees to measure the efficiency of outcomes at the expense of media quality. In that case, they are likely to receive the cheapest outcomes for the media seller to produce – outcomes that may have happened anyway. After all, with enough data it becomes easier to predict who will do something than actually influence their behavior.
Let’s not ignore the elephant in the room – large scale attribution requires a socially unacceptable amount of tracking. The real push backward would be if our industry decided more tracking of consumers and less transparency of media quality was a good idea at Facebook’s suggestion.
From finance to commodities to real estate, most markets trade on metrics that reflect the quality of the goods, with spot checks on outcomes. So it’s no surprise that advertisers are starting to hold publishers accountable to media quality standards based on attention metrics, while optimizing their own strategies to more efficient outcomes.
At Adelaide, we’ve found that a careful curation of metrics that signal the likelihood of attention is far more accurate at predicting outcomes than media currencies that treat all viewable impressions as equal.
One solution for consistent media quality measurement is a synthesis of Lumen and Facebook’s positions – an algorithm informed by research into attention metrics, using signals that predict attention to media and trained to ensure as close a proxy for outcomes as possible with opt-in data.
As Charlie Munger says “Show me the incentives, and I will show you the outcomes.” A seller’s incentive is to claim as much credit as possible, it’s vital that buyers don’t let them dictate outcomes in any sense.
Adelaide brings fairness and transparency to a digital marketplace damaged by easily gamed metrics. It’s our mission to quantify the true value of media, so marketers can invest more confidently and create incentives in the market for high quality consumer attention.