In preparation for their Omni.Digital conference in September, AdExchanger had one question for Tapad CEO and Founder Are Traasdahl:
Why and how is probabilistic relevant in a world populated by 800-pound deterministic gorillas?
Given space limitations, an abbreviated version of his response appeared in their excellent piece, live now. We thought you'd like to see the full answer Are gave to the question we are most often asked. Here it is:
"We believe a combined approach - using both deterministic and probabilistic - solves for accuracy and scale; and critically, it preserves an open ecosystem. There are three reasons probabilistic has an important role in cross-device solutions:
#1) OPENNESS: Our industry consists of large brands, large agencies, marketing cloud companies, marketing tech companies, ad tech companies etc. It is a vibrant ecosystem, each participant in need of a common cross-screen identity technology.
Take this example. We have a large auto client. They have an entire wall of data visualizations at their headquarters, depicting every customer touchpoint. It reveals that every single customer is touched by nearly 300 different marketing systems -- everything from websites, mobile apps and paid media buys to CRM systems, earned media and past-purchase records. A holistic strategy to connect those dots in a unified way is critically needed.
The fully deterministic players are not open and do not share their data with the rest of the eco-system.
#2 SCALE: Conservatively, we assume 180 million people in the US are on two or more devices. Of course, many of those people logged-in are on social media, but more likely than not, they are posting from just one of their devices with 75 other tabs simultaneously open. This doesn’t even factor in traditional TV -- which we’ve recently added to our capabilities in measurement -- which has almost no social log-in.
Given this reality, I don’t believe any company has cross-screen, log-in data for that many consumers - and definitely not from within their own ecosystem. It’s more likely that they can reach 40 - 50 million people across screens; and for most marketers, that’s still not enough.
Bottom line, scale becomes increasingly meaningful the closer you get to the 180 million number; it’s hard to see how any player in deterministic gets there.
#3 ACCURACY: The common wisdom in the industry is that deterministic data is 100% accurate. However, we see it ourselves (and hear it from social companies with deterministic data) that the maximum accuracy for any deterministic set is 93-94%. There are a lot of factors -- like shared devices -- that influence this number. We had Nielsen (and several social networks) confirm our accuracy at 91-92%. Yes, a few percentage points under a full deterministic dataset, but with notable advantages -- an open ecosystem, un-reliant on PII, the ability to deploy across all systems, and of course, greater scale.
True cross-device insight requires a combined approach, where algorithms predict connections based on probability, and then test accuracy with privacy-safe deterministic data. This gets you to very high accuracy levels, achieves scale and, when done right, fosters a healthy, thriving market for the entire industry."