Identity Graph Analysis at Scale Provides Ad Tech Agency Customers with Greater ROI
Challenge
When the agency was founded in 2011, they analyzed public declarations of location on
social media. As the company grew, that analysis expanded to other forms of declarative
data beyond social media.
But as time went on, people were not only using more devices, they were performing
different actions on each. And the Ad Tech agency’s original product wasn’t optimized for
recognizing a consumer across all devices.
“We needed to be able to capture data from multiple devices and associate it to a single
individual,” said the Chief Technology Officer of the Ad Tech Agency.
Even more challenging was the fact that pieces of data were continually changing, a result
of the fact that people frequently purchase new devices and clear cookies from their
browser.
As a company that defines and sells audiences to large companies across several
industries – from automotive to beauty to entertainment – the agency needed to find a
way to provide cross-device insights, fast.
They found their solution in Neo4j.
Solution
The Ad Tech Agency now relies on a stack that includes Neo4j, Node.js, Ruby, Go, Python,
Hadoop, Apache Spark, BigQuery and MongoDB.
When the agency’s identity graph receives a signal from a device, it collects an originating
device and/or cookie ID, which then points to the owner of the device – the user.
The identity graph then queries the device and all signals associated with it, and performs
processing based on the user ID. They can then export audience data to any target device
ID in the identity graph to potential advertisers.
Their identity graph relies on Spirograph (an internal Ruby app) to communicate with
Neo4j by either inserting data into or pulling data from the graph in Ruby, which is then
processed in Hadoop, Spark and BigQuery. The identity graph also relies on Cerebro to
convert user coordinates into a commercial