Consumers browsing online are used to ads following them as they move from site to site. But what about ads following them from device to device? So-called “cross device advertising” is set to explode over the next year as specialized data collection companies become ever more successful matching multiple devices to a single user, and advertising agencies become more sophisticated in how they use the technology.
In most cases to date, cross device ad targeting is fairly basic. A consumer who purchases from a company might see ads from that company on their phone, PC and tablet. But Steven Glanz, CEO of Crosswise, a provider of device-matching data, considers that user experience to be a failure of cross device implementation.
“If they were more intelligent about using it, you might really feel differently about the company,” said Glanz, who’s based in Israel. “If you made the purchase on your desktop and then saw a few ads on your mobile devices saying, ‘Thanks for becoming a customer. We hope you enjoyed the experience with us.’ That is less intrusive, more intelligent. They’re showing they understand who you are and what you’ve done.”
The need for such cross-device advertising is becoming more urgent as consumers change the way they search and buy online. In May, Google announced that “more Google searches take place on mobile devices than on computers in 10 countries, including the US and Japan.” But while searching has moved to mobile devices, consumers return to desktops when it’s time to buy, according to MIT’s Technology Review. Additionally, ad agencies have begun integrating video, mobile and display, making them more nimble at offering full-circle advertising.
Crosswise is a rapidly growing startup focused exclusively on providing device-matching data. The company uses big data, machine learning and proprietary algorithms to build an extraordinarily comprehensive and accurate device graph. It is one of a handful of companies worldwide working to link a mobile phone, desktop, tablet and television to a single user – without using any personally identifiable information.
The company collects more than a petabyte (a million gigabytes) of data every month on billions – yes, billions – of different devices. “We have a process where we extract features from all this device data we get,” Glanz explained. The data collected includes the IP address and the Wi-Fi network a device is accessing, GPS location, the device type, what browser is being used and actual browsing data, such as the sites the person viewed. There are about 30 such data points collected, and they differ from device to device. No personal data is collected.
This so-called “feature extraction” allows Crosswise to cluster devices so it knows there is a possibility that certain sets of devices are being used by the same person. Then there is a proprietary matching process that occurs. “Most of the secret sauce is in the matching process,” Glanz said.
“But companies that do not incorporate cross device matching data are definitely lacking in a big way.”
The result of the matching process is a device map – which shows the probability that a set of devices is used by the same person. That allows Crosswise clients to make advertising plans. “Our clients know if they have two devices that are 90 percent certain to belong to the same person, they will do certain things with that data that they would not do with data where the match is only 50 percent certain,” Glanz said. “In other cases, 50 percent certainty is more than enough.”
Crosswise has one of the largest teams in the industry. “What we do is at the very cutting edge of big data development,” Glanz said. “We have a lot of very smart people working for us. (Device matching) is extremely, extremely difficult.” Part of what makes it difficult is that the data changes dramatically from week to week, which means Crosswise updates its clients weekly, too. No one knows yet exactly why the data changes so dramatically, Glanz said.
While many marketers boast of an individual company’s data “accuracy,” it’s a misleading measurement. Data collection companies can tout accuracy rates of 90-plus percent. However, if the sample used includes only the data they are most certain of, it becomes a meaningless figure, because the sample is so small. A better set of metrics is “recall and precision.” Precision is the percentage of a device map’s matches which are correct matches. Recall, sometimes referred to as “reach,” is the percentage of actual matches in the covered region which are correctly matched in the device map.
The field remains an emerging one, but it is garnering a lot more attention. “It’s still a minority of companies that are using it,” Glanz said. “But companies that do not incorporate cross device matching data are definitely lacking in a big way.”
He said consumers so far don’t seem overly worried about the tracking involved. “Personal retargeting is something that freaked out a lot of people when it started becoming really widespread,” he said. “But no one is surprised anymore. It’s going to be the same thing with cross device. You’re used to seeing it within your browser, and you’re going to get used to seeing it across your devices.”
Retargeting is a particularly valuable application for cross-device data because for many types of purchases, “if the brand does not recapture that customer within hours or days, they’re probably not going to get the sale,” Glanz said.
“One of the big challenges with customer data is getting what we call the single customer view,” Glanz said. “You want to be able to link all your disparate systems so that you know for a given customer, when he’s been in touch with customer support, when he’s bought online, when he’s bought retail in a store,” and can target ads accordingly.
For example, if a retailer’s targeting system is not linked to the purchasing system, the company won’t know that a customer has recently bought an item. “They’re going to be bothering you with ads for something you just bought and probably don’t need again right away,” Glanz said.
Currently, it’s mostly data companies themselves that are using cross device data matching. “More and more large content sites and retail brands are starting to integrate cross-device data into their systems,” Glanz said. “There’s a lot of opportunity for adopting this sort of information.”
The amount of customer data available is unprecedented. Figuring out how to use it is still being worked out. “There is a lot of room for growth in how this data is used,” Glanz said. “Technology is one thing. Using the technology is another thing.”