In a recent dispute about the true cost of advertising, Lumen Research found itself at odds with Facebook. While Lumen envisaged how attention could be measured in the future, Facebook argued that its predictive algorithms are more precise at capturing attention in the present.
Both companies have ended up lost in translation. At the heart of the dispute is a "known unknown": we know that our senses and emotions matter to advertisers more than ever before; we just don’t know how to measure them at scale, yet.
Lumen recently put forward a new framework for measuring the "True Cost of Advertising". Its central thesis is that:
- All attention is visual.
- Visual attention to ads can be measured at scale, therefore visual attention could be treated as a commodity and priced accordingly. This would improve "price discovery", helping buyers know what they’re buying (and sellers know what they’re selling) and pruning unwanted "deadwood" along the way.
Facebook’s Ian Edwards and Harry Davison waded in with their response: "Understanding the true value of advertising attention". They postulate:
- Ad pricing should consider the outcomes that advertisers achieve – the end itself, not the means to the end alone.
- Personalisation enabled predictive algorithms are more efficient in gathering visual attention.
- Time alone is not a good proxy for valuing attention when outcomes are the primary metric for ad performance.
A solution in search of a problem
The debate hinges on some simple arithmetic to establish effectiveness: dividing a numerator (time spent) over the denominator (quality of attention) and normalising across different forms of visual advertising.
(The chapter title reminds us that only the English language uses the monetary term "paying" when describing attention – in Spanish we "lend"; in French we "make"; in Swedish we "give", for example).
There are three ways productivity can grow: by either getting more from the same, the same from less or more from less. The point being, one size doesn’t fit all when it comes to optimising the productivity of attention. Edwards and Davison point out that as Facebook allows you to measure (or optimise) productivity with 12 objectives. Video views (the sole focus on Lumen) is just one; brand awareness, app installs and conversions are examples of others.
Lumen’s narrowly defined one-size-fits-all proposal falls foul of appearing like a solution in search of a problem. At a high level, productivity tells us that we need to pivot in our desired outputs.
In brand advertising, where people are "out of market", you would optimise for less-from-more subtle impacts (making people feel they need to buy new running shoes); whereas in direct advertising when people are "in market", you need to achieve more-from-less (creating the urgency to actually buy a specific pair of running shoes).
A one-size-fits-all strategy of measuring time alone cannot account for these subtle yet distinct outcomes that visual ads can achieve. We need to consider connections the ads make, and these are not necessarily time dependent.
Connections would be a truer measure of attention, and yet...
Let’s revisit Donald Rumsfeld’s "known unknowns". Some ads form a strong connection, whereas others don’t, regardless of time spent – we just don’t know which half is being wasted, as Wanamaker famously said. To help answer this perennial question, I have written about the need to think about the trade-offs between relative importance and relative attention given to media.
For a recent lockdown example, binge-watching a documentary about tigers in captivity may take up a lot of time, but it’s not really that important. It didn’t form a connection.
On the other hand, ad formats that seamlessly merge into the content, like product placement on Hulu and YouTube, feel more like entertainment and less like interruptions and therefore form stronger connections. Yet, with the current tools and methods available, measuring the strength of connections at scale is next to impossible.
The challenge we face with measuring visual attention today has been the same since the invention of newspapers: there’s no closed loop between where the users’ eyeballs are and the content they might be looking at. We should applaud Lumen’s research as it warns against complacency and motivates us to define what the ideal solution would entail. After all, trusting human recall and manual diaries for measuring radio and television in 2021 is simply not fit for purpose.
This is where I want to share a lesson that a former researcher at Spotify, Sohit Karol, sketched out to me: “The Visual Attention Pyramid” of eyes, brains and hearts.
If we could finally "close the loop", this pyramid of eyes, brain and heart would be how ad effectiveness would be measured at scale.
At the bottom of the pyramid are the eyes. Most ads, while they might capture the attention of the audience, might result in low recall and low connection. Such ads provide lowest ROI outcomes, remain at the foot of the pyramid and should cost the least. Banner ad formats fall under this category.
In the middle of the pyramid are the ads that might capture the brain and result in recognition and recall. Repetition is one strategy to capture the brain, like the popular Geico tagline that everyone in the US is aware of: "Fifteen minutes could save you 15% or more on car insurance". While such ads might build recognition, there is no guarantee that they will form a strong connection. The cost of such ads should take this limitation into account.
The most effective ads will capture the heart at the top of the pyramid (and price). This happens when users not only pay attention to the ad, the ad experience becomes so seamless and entertaining that watching the ad and acting on the call to action itself becomes a value proposition. Many new social ad formats are increasingly falling under this category – form a deep connection and providing the highest return on investment.
By going beyond simple time-spent metrics, this pyramid would have profound implications for ad auctions. At present, auctions are a price discovery mechanism based on the formats, length and campaign objectives. In the future, auctions could discover the true price of connections.
The old vine of cookies can’t keep us off the jungle floor for much longer, and a reluctance to reach out to the new vines that aspire to a more refined and private assessment of human experience risks the old vine letting go of you.
Will Page is the former chief economist at Spotify and the author of Tarzan Economics: Eight Principles in Pivoting Through Disruption, published by Simon & Schuster on 1 April