“Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Thierry Cornet, partner portfolio manager at UM.
The market is awash with mobile ad networks claiming to be specialists in geolocation and, more recently, device bridging.
But pulling back the curtain can reveal an unpleasant surprise: There is often more sales spin than actual technological capability in these categories.
It can be very difficult to separate the wheat from the chaff in this space, but there are some specific points on each technology that should be part of every buyer’s conversation.
The prospect of being able to identify a unique user on their tablet, phone and desktop computer, and to know that an individual is linked to each of these devices, is a compelling proposition.
There’s a solid body of research that shows the benefits of cross-device exposure, especially for branding objectives. Companies in this space claim to use an individual’s unique attributes to “bridge” their various devices and attribute them all to the same user. This is commonly done using in a deterministic fashion using login data, such as mail service linked across devices, or probabilistically, using external predictive signals, such as a Wi-Fi router that serves several devices.
While deterministic methodology isn’t bulletproof, it does tend to produce more accurate results compared to probabilistically using technology based on statistics. Although probabilistic methodology claims to offer greater scale of matched devices, this isn’t always true.
Many of the companies in this space cite impressive “accuracy rates,” but keep in mind that the accuracy rate has nothing to do with the actual number of users exposed on multiple devices. Instead it simply measures the likelihood that the devices on which the exposures occurred belonged to the same person. For this reason, it's a good idea to look to partners that are third-party-verified by companies such as Nielsen or comScore.
One core factor critical to the success of a publisher matching devices at scale is how much PC inventory they have, which is usually ample, plus the amount of mobile inventory they have, which is often sparse. This is a major contributor to the low reach of unique users across devices that can be found in these campaigns.
For campaigns where device bridging is a consideration, media buyers need to be sure to negotiate for a solid percentage of unique users who will be exposed on more than one device upfront. Buyers can’t assume that just because they’re using a device bridging specialist, the majority of those reached will be bridged users; they need to get an upfront guarantee.
Location accuracy was a hot topic in 2014, and although it may no longer be in the limelight, the accuracy issues remain prevalent in the marketplace today.
The IAB issued a good set of questions [PDF] that every buyer should ask a geolocation targeting provider. The high-level conclusion one will reach after digging into the various location methodologies is that the most accurate data that you can buy at scale today is GPS-based data. The tricky part is that there are several ad networks that promise to tightly target whatever location you want, including something as specific as a storefront or parking lot. The reality, however, is often much different.
Various geolocation vendors may tell you that they have an accuracy rate out to several decimals, or are accurate within a certain number of meters, all of which sound good but can be confusing. What isn’t discussed is that those accuracy figures are often theoretical at best; they represent the absolute peak of what the technology can potentially deliver.
You can prove this point by looking at your location history on your own smartphone. This information is GPS-based, with Wi-Fi, so it’s the best data you can get without having beacons deployed. As you can see from the examples below, a location can span several city blocks worth of storefronts, which is far from the promise of being able to reach a very specific audience.
For this reason, buyers need to approach geofencing with caution. Advertisers are very likely going to have a high degree of inaccuracy, especially in denser urban areas.
Buyers should rethink strategies through several lenses. Would they purchase a billboard outside a specific location and would they be OK with targeting users who merely have the opportunity to see a billboard? That’s about as accurate as an advertiser can hope for with any scale.
Using an "audit" partner, be it qualitative via panels that allow their locations to be tracked, or quantitative with location tags that verify the precision and accuracy of latitude and longitude data within 100 meters, can also be helpful to advertisers. These partners can help ensure their location investment is at least appearing in the right general area.