28 Sep

Don’t Be Fooled, You Can Do Cross-Device Too…

Read on Screen6.io

What a week!

AdWeek is in full swing in New York City and cross-device continues to be one of the hot topics. From the CIMM’s release of the Best Practices in Cross-Device and Cross-Channel Identity Measurement to Google’s announcement on its new omni-channel products the industry is certainly experiencing sensory overload given all the new information, buzzwords and opportunities that cross-device places at their fingertips.

But, what does it all mean? Can your business benefit from it? Are you limited to just the large walled gardens if you want access to adtech’s shiny new feature? Let’s tackle these questions one-by-one.

– Accuracy, Precision and Scale (oh, my!)

Cross-device can be boiled down to a prediction that device A is likely to be the same user as device B. The effectiveness of any cross-device offering at making these predictions is a combination of three factors: accuracy, precision and scale. Accuracy is the percentage of all predictions that were correct. Precision is the percentage of all verifiable predictions that were correct. If all predictions can be verified, then accuracy and precision are the same. However, that’s where scale becomes so important. Often only a small amount of deterministic data is available, making the amount of verifiable predictions lower than the total of all predictions.

– The Unspoken Fallacies of Deterministic Data

Many in the industry view deterministic data as infallible, but it, just like probabilistic methodologies faces challenges. As shared by Oracle in a recent Adexchanger piece, “bad data issues are common… As many as 15% of emails in industry databases link with two or more households. This means that for many databases, this deterministic data is only 85% accurate.”

Moreover, the industries heavy reliance on log-in data for both targeting and verification is ignorant. How many of you have more than one email address? Each of those accounts have different cookies associated with them. For example: Cookie X is tied to Keith@Mail.com; and Cookie Y is tied to Keith@Screen6.io. With a probabilistic solution, like Screen6, these two cookies will be predicted to belong to the same individual, but a deterministic dataset would split these cookies and classify them as separate identities.

Even the largest publishers, like Facebook, with logged-in user bases have challenges with scale and must supplement with the inclusion of a probabilistic method to understand their entire ecosystems’ activities. And Facebook has the additional challenge that a significant percentage of their user base mostly uses Facebook from their phones. As a result, they have massive scale in mobile, but a much smaller footprint on desktops.

Oracle goes on to make some very strong points focusing on building a strong probabilistic methodology by ensuring the use of clean data, a strong algorithm, and verifying the base deterministic dataset. However, the pressing and underlying issues specific to cross-device are actually data rights, data storage, and local privacy laws in countries across the globe.

– Privacy Safe

It’s funny how myopic we’ve become when it comes to privacy. The Orwellian nightmare around privacy in 1984 was the notion that one massive all-knowing entity eventually rose to control people’s lives. Being “Privacy safe” doesn’t just mean adhering to law and industry privacy standards. Privacy safe means being very careful as an industry not centralize too much data with any one company granting them the right to normalize across all data assets. Privacy safe means refraining from incorporating PII so that all this data doesn’t live on forever.

As an industry, we must find a solution where the unifying platform is not in the data business itself. The entity should not own any data, nor house any PII itself. Lastly, privacy safe means being really clear regarding what data gets transfers and to whom. These challenges increase exponentially as the device graphs scales.

At the end of the day, after sifting through the layers of marketing fluff advertisers will hopefully see what each platform truly offers. The understanding that Sizmek can offer cross-device frequency capping and Audience Science can leverage mobile profiles for desktop media buys and AdForm can attribute a conversion on one browser with a prior impression on another – these are the realities, and Screen6 is the cog in each of these forward thinking technology stacks which enable the cross-device capabilities. We do so in a privacy compliant manner, at scale, with consistent precision, without violating privacy or data rights.