Location-targeted mobile ad sales are expected to rise from $9.8 billion in 2015 to $29.5 billion by the end of 2019 — but with that rising demand comes greater expectations about ROI, attribution, and accuracy.
The issue of geo-data accuracy has always been a thorn within the great promise of location-based advertising to bridge online and offline, desktop and mobile (and we first outlined the problem here back in 2014).
Placed, which was acquired by Snap this summer to provide attribution services for Snapchat ad clients, is the latest geo-data specialist to offer guidance on the matter of location analytics accuracy in a report, Accuracy & Bias In Ad Exchange-Derived Location Data.
The analysis for the report is gleaned from the company’s primary product, Placed Attribution, which is based on an audience of over 150 million device generated over 140 billion latitudes and longitudes on a monthly basis.
Thanks to its extensive alliances with dozens publishers, networks, and demand side platforms, including ncluding IPG Mediabrands, DigitasLBi, Horizon Media, Tapad, DataXu, Drawbridge, among others, Placed’s data sources have grown to include first and third party data as well.
As Placed CEO David Shim describes it, location ad accuracy is a viewability problem.
- Average location accuracy was 4 New York City blocks
- Only 1 percent of locations were accurate enough to identify a store visit
- 80 percent of bid requests with location occur when they are in-between visits, with a good portion of the visit based impressions occurring at home
“With location accuracy not in the forefront of the viewability conversation, bad players have been able to capitalize,” Shim said. “Placed’s recent research and findings arm advertisers with an understanding of the location landscape that has been missing to date, independent of media.”
The Viewability Issue
In the larger ad tech sense, viewability has been a considerable cause of mistrust between buyers (brand and agencies) and publishers. In essence, viewability refers to whether a display ad placement was seen by an actual person or if it fraudulent ad impressions were generated by the “visits” of bots.
“In the near future, marketers will require a viewability-like metric to gauge the accuracy of location used for media, targeting, attribution and analytics,” said Benjamin Bring, VP, Mobile Media Director at Ansible, in the Placed report. “To date, the siren’s song around the potential of location has been able to drive the early adoption, but scale won’t come until the industry delivers a standard verification solution.”
“Viewability in location is directly aligned with accuracy,” Shim told GeoMarketing. “In an ecosystem today, where anyone can claim any location, it is important to not take location at face value, and continually verify the source of the location.”
‘Not All Geo-Data Is Created Equal’
Among the most difficult issues for marketers who want to use location data comes down to the sourcing of that information. Those sources, or signals, can come from a variety of channels, including GPS/satellite, wifi, computer IP-addresses, cell phone towers when it comes to pinpointing the specific lat/long the device accessed. Panel-based check-in services like Placed’s— the location-based ad equivalent of a Nielsen diary that contains what a viewer watched on TV — are another popular avenue for accessing location data.
As programmatic advertising has become mainstream, the general purpose for for location advertising is two-fold: there’s the desire to provide real-time ad targeting as well as developing a greater understanding of consumers according to the places they go that provides more actionable insights than mere demographics (age, gender, household income, etc…) can offer.
One of the problems with a source like bidstream data is that its not a persistent signal like wifi or GPS. Bidstream data often depends on a person opening an ad on their phone while they’re in a specific place. The publisher whose ad is opened in that moment receives the data and passes it on to the network or vendor that placed the ad. If that person who saw the placement then goes to a store that was advertised, that visit counts as being “attributed.”
Of course, a phone’s location services records signals from hundreds of places in a given day. The odds of the information being attributed coincidentally (i.e., incorrectly) is a challenge that comes with real-time data platforms.
Quality bidstream location data, like those coming directly from publishers, is generated from the mobile device native location-based services, which use a combination of GPS, wifi, and other signals. That data is then pulled by the app developer via the phone’s SDKs. The process is the same whether the app sends location data via the exchange (bidstream) or direct. Regardless of a location data’s origin, bidstream or direct from publishers, it’s important to filter data and curate sources as well as recognize and filter low-precision signals.
“Not all data is created equal,” said Joao Machado, director of Mobile at OMD, in the Placed report. “You have to be very diligent in determining the accuracy of location data coming off the exchanges.
“Quality beats scale all day long and 1st party data is the gold standard in quality of location data,”Machado added. “A horoscope app dumping location data into an SSP before landing in the exchanges is the example of what leads all of us to scratch our heads around lack of value around so much of what goes on in the market.”
As we noted above, the average location accuracy Placed found was within four blocks — and only 1 percent were able to identify a store visit. That all sounds pretty dismal and much less reliable than the image a user has when looking at their own location on a map and seeing the accuracy within 3- to 20 feet on average, according to industry sources we’ve spoken to.
While it’s a given that location accuracy is, all over the map (pardon the pun), is this a matter of the difference of location signal sources, such as GPS versus cell towers versus wifi versus bidstream data?
“It’s a combination of factors,” Shim said. “When measuring directly from the device, location signal can vary based on a number factors include environment. This said, the primary source of location inaccuracy is bad players on the exchange that treat cell tower location the same as GPS, map IP address to latitude and longitude, or commit outright fraud by applying a latitude and longitude to a non-location based impression.”