In part two of our “Debunking Myths in Location Data” series, we explore accuracy and ad fraud.
Myth: Too much of the location data in the market is fake.
Data vendors often try to supplement the limited data they own with data sets from their competitors and other, unknown sources.
As a processor of location data, we deal with this problem every day. Location data varies in quality – from clean data to outright fraud. However, rather than throw the baby out with the bathwater, we’ve developed ways to filter out fraudulent or suspicious data.
At Gravy, we analyze each and every location signal individually, taking into account all other recent location signals for the device, as well as all signals for devices in close proximity to the signal being evaluated. This allows us to see patterns, and to differentiate between expected behavior and anomalies.
Gravy’s unique AdmitOne processing engine and Forensics technology classifies and interprets location data – while filtering out problematic signals – to support accurate analytics at scale. In fact, each location signal that is processed in our system is assigned a forensic flag – such as “cell tower derived”, “fraudulent”, or “unexpectedly high signal count”.
Myth: Bid stream data is particularly problematic.
Up to 80 percent – or more – of bidstream location data available is fake.
Again, while there is definitely fraudulent data, there are ways to identify it, filter it out, and use only data that has been verified.
Gravy recently analyzed a source of bid stream data for fraud. While we did find about 30 percent of bids to be suspect, we also found that not all apps generating bid requests are created equal. While some apps generate 99 percent of signals fraudulently, others exhibit no fraudulent behavior. We also found that the rate of fraud greatly depends on the ad exchange.
Myth: We don’t even know that the data works.
People don’t just walk into stores with their phones in-hand using an app or surfing the web; that’s not how people behave when shopping. There are also unscrupulous players littering the ad exchanges with fraudulent data so that advertisers will think someone visited a store after seeing their ad.
Some people *do* seem to be on their phones all the time and everywhere. It’s also worth noting that location data supports many use cases beyond advertising and attribution – like competitive intelligence, CRM enrichment (powering advanced personalization and loyalty programs), and fraud detection.
More importantly, as outlined above, the key to finding data that works is good quality location data that’s been processed using sophisticated algorithms to filter out any spoofed or suspicious data. This is exactly what Gravy Analytics does.
Finally, let’s not overlook the fact that deterministic data
includes location information. When compared to other methods, physical world visits are the strongest possible indicator of consumer interests and affinities. To illustrate, I might research a Ferrari online, and cookie data will show as much, but I am highly unlikely to ever buy a Ferrari. In contrast, if I attend Andrea Bocelli’s upcoming concert at Madison Square Garden, it’s almost certain I’m an opera fan. As we say, “Where we go is who we are.”
For more information on how Gravy Analytics ensures quality and accurate insights, visit us here