What if everything we thought we knew about advertising online was wrong?
Over the past year, pressure and regulation on digital ad platforms has ramped up to the point that our ability as advertisers to target people based on their behaviour is down significantly, but is that a bad thing?
The promise of digital advertising, since its inception, has been the ability to micro-target messages to specific audiences based on what they like, where they live, and what they’re up to online. An entire generation of media buyers have worshipped at the altar of behavioural targeting, and as it’s being methodically pulled away from us, it is revealing some clues about a question that we should have been asking all along: Does targeting people based on their behaviour make our results better?
It seems like it should be a given, right? If I’m a travel destination with lots of outdoor adventure activities, I should want to reach out to people who are frequent travellers, who like to hike, surf, or ski. I may even choose to refine my audience based on the number of children that they have, or their marital status. Intuitively it makes sense, and the data shows that certain behavioural targeting options will sometimes increase effectiveness.
Let’s look at that same relationship from another perspective: one person in that audience. She’s a 35-year-old woman who lives just outside of a major city in the Pacific Northwest. She regularly engages with Arc’teryx content and takes trips three times a year. She makes a good income and has a family that she likes to take into the outdoors with her.
She seems like an ideal candidate, so according to behavioural advertising theory, our destination marketer would do well to find a group of people like her and hammer ads at them. But our example isn’t always thinking about travel. Sometimes she’s researching career options, or shopping for a birthday present for her friend. There are times when she opens up YouTube to fantasize about her next trips, but that’s only one part of what makes up her personality and attention, so the ads that found her when she was doing just about anything else fell flat.
Now, let’s consider that same destination ad strategy through a different lens. The theory of contextual advertising is that what matters most to an ad impression is what’s happening around it. Its placements are targeted based on the content of the website, blog post, or video that the ad appears within. A low-tech version of contextual advertising is when a movie, podcast, or an influencer pitches a product within its content. There is no dynamic audience targeting going on there – the people who see the pitch are simply the people who are consuming the content.
In this scenario, our audience of outdoor travellers wouldn’t be targeted with ads across their entire web experience, but if the ads are well planned, the potential visitors will find them when they are reading travel blogs, or watching a YouTube video about adventures. The content would find them at a time when they are most receptive to the messaging. On a per-person basis, it is much more difficult to create and track multiple impressions, but it is very likely that the impressions that we do get are being viewed by an audience at the right time when they’re in the right mindset.
Of course, that sort of manual placement doesn’t scale, and as ad budgets get very large, or very small, it’s not realistic to think that the marketers responsible can be reaching out to media platforms to negotiate direct placements. Instead, technology is making it possible for ad networks to read and index the content in a page, post, or video and automatically add that placement to a database of options. Then, much like current ad networks pick and place our ads in front of targeted users, contextual ad networks place our ads within relevant content, allowing the users to find the ads for themselves.
Google CEO Sundar Pichai says the company hopes to bring the contextual advertising that Google search is known for, to YouTube. Sees big opportunity in that space.
— Alex Weprin (@alexweprin) February 3, 2020
This idea is not new
Google AdSense has placed ads directly on websites for nearly two decades, and advertorial content has been placed into newspapers for over a century. What’s changed is that, as behavioural tech grew and was pitched by the big social networks, we didn’t stop to ask whether it really is more effective. What if all of the work that we’ve been doing over the past decade has been based on the flawed assumption that people are simple, static versions of themselves? What if, instead, we were all complex, flawed humans with illogical wrinkles to our personalities and preferences that shift as frequently as we want them to? If that’s true, then no amount of targeting based on past behaviour, or placements based on our Facebook likes can predict what one person, let alone an entire cohort of people, will do in the future.
Fortunately, you don’t just have to take my word for it. In a 2018 FTC hearing, a Carnegie Mellon University professor shared the results of his study where he worked with a major publisher to compare the results when they used advanced behaviour-based targeting versus ads that used no targeting at all. The sample size was millions of transactions, and the results showed that after all of the tech and labour that went into the behaviour-based ads, they showed only a 4% increase in revenue per ad.
4% is not nothing. 4% could help you hit your quarterly goals, or it could flip a negative return on ad spend to positive. The hook, however, is the rate that those ads were going for. On average, the behaviour-based ads were 500% more expensive, meaning that even after the 4% increase in effectiveness, they were still paying over 4.5 times as much per new customer acquired.
Publishers are also experimenting with the removal of tracking data. The New York Times’ Europe office cut off cookie-based tracking, which was supposed to cause advertisers to run for the hills, but instead, they increased their ad revenues because the media buyers now had more control over where their ads appeared.
Don’t throw the baby out with the bath water
There is no question that there is still a place for behavioural data on a micro-level. Retargeting to past guests, and abandoned cart ads are still incredibly effective, but the way that the privacy laws are trending, I don’t recommend betting that even those will still be available to us for much longer.
Behavioural advertising isn't paying publishers. It's not what brings you free stuff on the Internet. It's just how Google and the rest of adtech sells your data. What value it does have – to advertisers – can be delivered without surveillance.
So let's. https://t.co/UigdwKBghB
— Robin Berjon (@robinberjon) May 30, 2019
Instead, the best advertisers in 2020 and beyond will reconsider the assumptions that we’ve all made to start thinking about where and how we’re showing up for our audiences, rather than just who we’re targeting.