Crosswise dug into the data to build and release Thursday a cross-device identification platform relying on statistical models and first-party data to improve ad targeting and measurement across devices.
The platform analyzes data and uses machine learning and proprietary algorithms to support ads driven through demand-side platforms, data management platforms, agency trading desks, affiliate marketers, analytics companies and attribution providers.
The technology links Internet-enabled devices to one user without tapping into personally identifiable information. Brands can use the information for a variety of business models across devices, including audience extension, targeting, frequency capping, sequencing, measurement and attribution.
Crosswise collects billions of non-personally identifiable data points like location data, domains browsed, apps used, age, gender, WiFi hotspots, shared ISPs, and many others. The company's scientists analyze the data and look for patterns to determine if specific phones, tablets and PCs are owned by the same person.
"To train and validate these statistical models, we have access to a massive set of deterministic data based on user login data," explains Steve Glanz, CEO and co-founder of Crosswise. "We've partnered with several companies to obtain the data for tens of millions of devices, so we know for sure that a specific phone and PC are owned by the same person. That data is critical for training and validating our statistical models."
To ensure the quality of the data, Crosswise offers brands a quality score based on the first party data in the deterministic data set.
The technology aims to solve the mishaps of cross-device identification. Marketers recognize that consumers browse and complete transactions on multiple devices. In an industry built on making decisions based on data, lack of understanding around cross-device identity can lead to making bad decisions and inaccurate results.
Glanz said that even the most basic functions of frequency capping and sequencing become impossible when running a campaign on multiple devices without knowing how those devices tie back to an actual person. It also can fix another broken model -- attribution -- because the technology helps marketers understand how to follow consumers who click on a PC ad and then convert three hours later on a tablet.