By now everyone is familiar with real-time bidding (RTB). RTB was a revolutionary development in digital advertising because it made it possible not only to target specific users as they moved across the Internet, but to do it as efficiently as possible via a real-time bidding process. But if it’s common knowledge that marketers can bid in real-time, what isn’t widely known is that those bids are often based on data that’s hours, or even days, old. In other words, RTB isn’t quite as real-time as you might think.
To appreciate the problem, you have to first take a step back and remember that the entire RTB process is driven by data gleaned from a user’s behavior. In its most basic form, site retargeting, RTB comes down to bidding to serve ads to users who have visited your site. But that’s just the beginning of what you can do with RTB today. Vendors now offer targeting based on intent revealed in searches, time of day, device, geographic location, items in a user’s shopping cart, and on and on.
Among other things, this means that audiences are identified in many different ways. You might be targeting all people who are shopping for a Ford car, or you might be targeting only specific people who have visited a specific page on Ford.com. Increasingly, “Big Data” is being used to identify audiences, and the reason is obvious enough: The more data you’re relying on to identify the audience you’re targeting, the more sophisticated your bidding algorithm becomes.
So far, so good. The problem is that working with Big Data requires a lot of process power. Right now, identifying an audience out of mountains of data often relies upon MapReduce, a programming model that makes it relatively easy to process large data sets. But MapReduce’s ease of use comes at a steep price. As noted above, it can sometimes take hours or more to process the data you need. And that delay is ultimately going to undermine your targeting.
Take the case of someone looking to book a vacation online. A certain percentage of users will purchase tickets only after a short period of browsing. If it takes three hours to know an individual is worth targeting, the tickets may have already been purchased. Worse, when the ad finally is served, it’s money down the drain. You’re serving an ad for a vacation package to someone who bought tickets an hour earlier.
The medical world provides a good analogy here. Doctors often talk about the concept of the “golden hour.” After an accident or trauma, if patients make it to a care unit within this critical first hour, their chances of surviving and fully recovering are hugely increased. Miss that golden hour and the patient outcome dramatically changes. The same goes for data. If you’re not able to serve an ad when you know someone is in buying mode, you’re missing the golden hour of marketing – and wasting a lot of money.
The good news is that an increasing number of companies have launched alternative technologies that can process data dramatically faster than MapReduce. With these new technologies, audiences can be segmented and acted upon only milliseconds after the data has been collected.
It’s often said that data wants to be free. But the rise of RTB is a good reminder that data also wants to be fresh. And if we want real-time bidding to live up to its name, we’ve got to make sure that it’s powered by real-time data.
Time is running out to feature your company in our inaugural Mobile Vendor Reader Survey.
Marketers create personas to better understand their target audience and what it looks like. If marketers can understand potential buyer behaviors, and where they spend their time online, then content can be targeted more effectively.
What’s behind a successful data-driven marketing strategy?
Audience targeting can be challenging in social media, especially when brands make quick assumptions about their target users. How can you avoid generalisation and what are the real benefits of it?