The rise of real-time bidding in recent years should come as no surprise to anyone familiar with the technology. Real-time bidding works because it allows the advertiser to cherry-pick a potential customer out of a pool of hundreds of millions. Once you have this tool at your disposal, the old way of serving display ads — buying a site's entire audience and hoping for the best — just doesn't make a whole lot of sense anymore.
In most cases, the targeting that powers real-time bidding is informed by a user’s online behavior. An advertiser might want to serve ads to someone who has visited the company’s site without converting (site retargeting) or to someone who has searched for a specific keyword (search retargeting).
These techniques are “reactionary” in the sense that the targeting is performed in reaction to something a user has done online. Reactionary targeting is incredibly powerful. By honing in on individual users based on what they actually do, Óonline advertising transformed from an art into something much closer to a science.
Though effective, there is one inherent limitation to reactionary targeting: you can only target individuals who have done something that reveals their interests. Or, put another way, you can only react to people who are lower in the funnel and have self-identified themselves as potential targets.
But what about all the people out there who haven’t visited your site or searched for the keywords you’re targeting, but who are nevertheless ideal targets for your brand? You simply can't afford to ignore these people, and that's why you can't afford to ignore another technique known as “predictive targeting.”
Predictive approaches, such as look-a-like targeting, can "predict" which user might respond to an ad based not on one specific piece of data, but rather on a wide range of different attributes. One of the best way to predict who will convert is to look at your existing converters.
Thanks to programmatic platforms, which include a universal profile on hundreds of millions of people, we can look at the shared attributes of your converters. Most attributes are not necessarily human-readable. For example, many converters might have all read the same very obscure four articles. This data isn’t necessarily just noise. It might reveal a pattern that allows us to predict that other people who have read those same four obscure articles will be especially likely to convert. Those individuals can then be targeted in a real-time exchange, in the same way a brand might target someone who visits its site.
Predictive targeting of this nature is extremely important because it allows you to find new customers, users who might otherwise forever remain off your radar. This isn’t to say that reactionary targeting can’t find new customers. While site retargeting is limited to people who have already visited your site, search retargeting can also be a fantastic tool for finding new customers. Take, for example, a cruise line that wants to server ads to individuals who have searched for "Caribbean cruises” on Google. Because these individuals haven’t yet visited the cruise line’s site, they’re new customers. Using search retargeting to serve display ads to them in the minutes/hours/days after their search can be extremely effective in driving conversions.
But, if there’s one thing most marketers can agree on, it’s that you don’t want to limit your pool of new customers. And once your search retargeting campaign brings in new customers, you’re now in a position to examine all of the data on those new customers to determine whether they share attributes with other users who you would want to target in a real-time exchange.
The power of predictive targeting, in other words, is that it locates key targets that you’ve been missing and gives you even more chances to bid for new customers. That’s why you’ll be seeing more and more of it in the years to come.