Where Is My Money Going? Examining the Problem of Bad Location Data

Approximately 59 percent of location data is inaccurate, writes Placecast CEO Alistair Goodman — but that doesn’t mean all targeting hope is lost.

The breakneck pace of innovation in mobile advertising has been driven by the emergence of mobile as the dominant device for media consumption. Only a few short years ago, we worried about the lack of cookies and “mobile blindness” on user data — but fast forward, and we now have a seemingly perfect view of the mobile consumer.

Location data has been a big driver of this innovation, enabling advertisers to connect digital and physical world behaviors. But the reality is, much of the location data that’s out there is not as accurate as promised: In fact, a recent Forrester study revealed that one-third of digital marketers in North America cited inaccurate location data as the leading challenge their organization faced.

This problem of inaccurate location data has been in the news more and more, and with good reason. A recent study found that 59 percent of location data is inaccurate — meaning more than half of the mobile ad impressions paid for by brands are not delivered within the promised target range. To make matters worse, iOS and Android updates will increasingly limit location data in apps, with only a handful of apps generating background data — making it even harder for marketers to target users’ proper location. And if location-centric messages are reaching the wrong audience, the message is useless and advertising dollars are wasted.

Adtech brands are noticing this, too: Snap’s recent acquisition of Placed and Zenly, for example, as well as its partnerships with measurement companies, show that marketers want to see accurate ROI when it comes to location targeting. In a world wrought with mobile ads, successful brands promise consumers a personalized experience — and accurate location data is the best way to deliver on this promise.

Hyperbole masks a real problem

Location-based mobile ad spend has seen a massive boom: According to BIA/Kelsey, nearly 40 percent of the $40B in U.S. mobile ad spend in 2017 will have a location targeting component. Agencies and brands are embracing the use of location data for everything: from planning out-of-home (“Who drives by that billboard”) to building audience segments for targeting (“Walmart shoppers, please”) to proximity ad campaigns (“Let me fire you an offer in front of my store”) and ad effectiveness measurement (“Who visited my store post-ad exposure”). These, and many other use cases, are all enabled by location data.

When taken as directional indicators, they represent a real breakthrough in marketing: the ability to close the loop on real-world behavior. However, media buyers are growing increasingly wary of these claims. According to a recent MMA study, the top 4 concerns of mobile ad buyers revolve around the quality of the data. Their suspicions are well-founded: Most of the location vendors have written extensively about the poor quality of location data signals and begun to put technology in place to screen out bad data. At the same time, the MRC and MMA are doing good work on emerging standards, but we’re still a long way away from a clean, consistently accurate signal.

The location space is the poster child for tech-enabled promises that capture the imagination of media creatives and buyers alike, with companies claiming to be able to locate a consumer standing on a postage stamp. However, the narrative about precision has missed the point: While it’s technically possible to precisely locate someone some of the time, getting consistently accurate data for large-scale audiences is much more difficult. In reality, we’re just beginning to work with this data at this scale and uncover its promise.

Therefore, constantly reinforcing the message that ad buyers can precisely target is a disservice — especially when media underperforms because it’s being delivered a long way from where it was intended.

So how big is the location data problem, really?

Depending on the source of the data, location inaccuracies vary from below 10% coming directly from a select set of apps and SDKs to 80 – 90 percent in the programmatic world. Placecast’s internal analysis (run by comparing location data to a truth set from carrier data) indicated wide variations from 8-92 percent inaccurate, while in programmatic exchanges, it can be almost 60 percent inaccurate. Removing the walled gardens from the equation, this means that as much as $4B of geo-targeted mobile advertising may be going far afield (pun intended). It’s not that the data says a consumer is across the street — it’s more likely that they are on the other side of the country.

There is a myriad of reasons why this happens, from the way the data is generated on the handset to how it is collected by apps and SDKs, and how it gets passed into the ad ecosystem. Bid prices for location inventory are at a significant premium, creating an incentive for bad actors to “improve” the location precision by substituting wildly inaccurate data. The harsh reality is that a significant portion of location targeted media is still not going anywhere near where it was intended. And, in our experience, using pattern recognition to filter out this data only captures about half of the bad data — but there may be another way.

Carrier data to the rescue?

As the mobile-location space continues to evolve, standards and best practices are beginning to emerge which, in the long run, will be good for everyone. We’re also beginning to see the emergence of new scale players in the form of mobile operators – Oath (AOL-Yahoo-Verizon), for instance, has offerings that are beginning to bring large-scale, deterministic, properly-consented mobile data sources to apply to media.

Carriers are the only entities that can deliver an all-encompassing view of location data via background apps, geotags, and location-enabled apps. While this data is usually not available for sale at the individual level, agencies and brands can increasingly leverage it in aggregate through emerging services like Oath. Because of this potential, the carriers become an intriguing new entrant with the capacity to vastly simplify what is a very complex mobile data space.

However, carriers also have to carefully manage the primary interests of their customers; as an ISP, they are stewards of a higher standard on privacy and data sharing. Tim Armstrong is espousing a future in which the carriers collaborate to create a third walled garden to rival Google and Facebook. While this is one possible future, it also relies on the belief that carriers, who compete viciously for new customers, will in fact happily collaborate on media. The track record on carrier collaborations is not great — but then again, the opportunity for new revenue has never been so large.

Another possible future is carriers exposing products and services through trusted partners, in a more loose federation. Regardless of the approach, what’s clear is that the carriers are stepping into media which can only mean good things for brands, given the high-quality of their signal data.