Hypertargeting at Scale

Thanks to AI, the future of mass media communications is 1:1

Over the last half-century, marketing has steadily progressed from mass to niche. From three television networks and forty million people watching I Love Lucy at once, to dozens of TV channels where three million viewers is a coup. The internet furthered the trend, with thousands of professional content sites sprouting up, each of whom would be lucky to have hundreds of thousands of daily users. And finally today, with social feeds so personalized that it's effectively a media channel tailored for one. 4.62 billion social media users worldwide… 4.62 billion unique streams of content.

This evolution has driven increasingly sophisticated targeting mechanisms. Once we used media planners to dissect just the right places and times to place media to ensure their audience sees a marketing message. Now we delegate it to algorithms that have become exceedingly good at finding just the right audience; ones that are receptive to our messages at just the right time.

But what has not changed, until very recently, is the advertising content. Ads that ran in the 1970's are not substantially different from those of today. Put aside the clothing they wear, many wouldn't look so out of place on Hulu. They're still a pithy story in 30 seconds, selling hopes and dreams made possible by the product du jour.

And yet the same 30 seconds to everyone is increasingly dated. Our media buys are hyper-targeted. Why can't content also be hyper-targeted?

The idea of hyper-targeting content is simple. If marketers can laser-in on just the right people to talk to, why can't they also deliver to those individuals just the right message? In today’s world, shouldn't there be a way to have messaging be truly 1:1, but done at scale?

The problem, historically, has been the difficulty inherent in content creation. It's expensive to shoot a fully-produced advertising spot. If the ad isn't going to be seen by a large number of people, it is hard to justify the investment. Anything approaching 1:1 is only possible if ads can be cheaply and rapidly produced, without a human sitting down figuring which person is going to get which message, and then spend even more time making it.

But three recent innovations have finally started to make hyper-targeted messaging possible, and effective:

Artificial Intelligence

Artificial intelligence has proven its worth in the marketing world when it comes to reaching target audiences. We've all seen how exceedingly good Facebook, TikTok, Google, and even discovery platforms like Taboola are at finding the right audience for any given message. But what's started to gain traction is the use of artificial intelligence to build the marketing message itself.

For example, Google's Responsive Search Ads can automatically generate and test many derivatives of search ads from a universe of content provided to them, allowing the marketer to automatically find the message that most resonates with different search queries and segments. Google takes a similar approach to its responsive display advertising product, adding images to the mix.

Other products similarly promise copy automation for email, ads, and landing pages, many of whom rely on OpenAI's GTP-3, which has shown to be fairly impressive at producing copy that feels surprisingly human. Crops of startups go beyond this, using AI to mix and match audio, video clips, and copy to mass-produce social ads that are ready for mass-testing. Pencil is one such example. Thanks to AI, it's now possible to produce high volume content to test which messages really work for which audiences.

Homegrown Content

Part of what makes AI content creation possible today is that the bar for what constitutes an acceptable ad, from a production-value standpoint, has never been lower. In fact, content that is raw, homegrown, and authentic often resonates more with consumers; therefore making it more effective.

That's because social hasn't only turned us all into content creators, it's also trained us to love the authenticity and spontaneity that social content uniquely affords. The ten most popular TikToks are decidedly homegrown. It is just a person, like you and me, doing something interesting in front of their phone. Really fun stuff. Also really cheap to make.

The savviest advertisers, seeing the appeal of homegrown fare, have followed suit. With homegrown content's low production costs comes an ability to produce substantially more content for the same amount of money. It comes with the added bonus of having content that feels more real, relatable, human...and therefore more persuasive. Just whip out the phone, start shooting, and edit away. Talent isn't even necessary; just have the people behind the products start talking or recruit existing customers to sing praise.

Marketers can go further on the road to hyper-targeting by skipping the creative process entirely, and outsourcing it to the niche the brand is trying to target. For every niche, a vibrant world of content creators exist, who would happily produce material that perfectly markets the product. No need to figure out the messages that work. And there is no shortage of brand fans who would happily talk about the products they love, even if they don't have much of an audience. Just build a community and engage. Then they'll produce, and all you have to do is amplify.

Progressive Creation

The low bar for what constitutes advertiser content means it's easy to test large numbers of messages to many audience niches, and then progressively invest in higher production value content for what really works. In short, use the platform's targeting algorithms to teach you what messages work and then spend the big bucks to produce ads that can have more of an impact.

In this regard, the savviest political campaigns have led the way. They'll start with a matrix of niche audiences, the political issues they want to strategically focus on, and the messages that could work for each segment. Simple ads can then be produced and run on Facebook to see what works--for example, just a photo with some text on it. The ones that win can be upgraded to a short video spot, relying on just a quick iPhone clip or some video pulled from stock. If that works, the possibility of a professional shoot opens up.

Then, the ad can be fully exposed to the niche, and derivatives can be tested to see if they'll work with other targeted audiences. In doing so, the ads can start to run on broader platforms like [gasp!] TV. At the same time, they'll try new basic photos and graphics to see what can beat the originals. Then try entirely new sets of messages, and move on to new issues as the current events dictate. Rinse and repeat.

Brands would do well to emulate.