Accuracy is a spectrum when it comes to location data.
And because there’s no systematic way to discern between different types of location data, including how granular it is or when it was derived, the buy side is confused and reluctant to spend.
“When we ask clients what the stumbling block is for location, they say, ‘It sounds great in theory – but we don’t know what we’re getting,’” said Michael Lieberman, COO North America at WPP agency tenthavenue. He’s also co-chair of the Mobile Marketing Association’s location data accuracy working group, which released a white paper in mid-October delving into the various factors impacting location data accuracy – or the lack thereof – and what can be done about it.
The MMA’s No. 1 takeaway: “Transparency is becoming the biggest driver of revenue,” Lieberman said, speaking at an MMA event on location data accuracy Wednesday night.
Seems simple enough. More transparency equals better quality data equals more dollars. But it’s actually a little more complex than that.
“The first step for us was to figure out why location is not actually driving the quality we all expected it to,” Lieberman said. “That sent us down a bit of a rabbit hole.”
All Over The Place
Up to 90% of the location data appended to ad inventory is inaccurate, according to Skyhook Wireless, and there are many legitimate reasons other than fraud for why that is.
Among them: truncated, transposed or mislabeled lat/long coordinates, prematurely interrupted data requests and apps that hard-code their GPS data and never update the signals they send into exchanges.
Another culprit is centroid processing, a method by which publishers derive location from a centralized IP address using a ZIP code or a designated market area as a proxy for a user’s real-time location. As Mike Schneider, VP of marketing at Skyhook Wireless, quipped, it’s a bit like inviting friends over for dinner to your apartment in New York City – without telling them your actual address.
A lot of these errors slip in because there are no official rules around quality control.
“Contrary to the prevalent thought out there, we found that a lot of these issue stem from a lack of standards and education and also because of technology reasons [rather than] fraudulent behavior or malicious intent,” said Vikas Gupta, VP of marketing at Factual.
Next Steps
The MMA is in early talks with the Media Rating Council (MRC) to develop official standards for location data transparency.
The goal: The industry must standardize how to measure offline foot traffic generated from mobile marketing, said MMA CEO Greg Stuart. This will bring “clarity into a space that is today burgeoning with varied methods and competing claims. Standardization benefits us all.”
But considering that the MRC is knee deep in mobile viewability right now, it’s unlikely that there will be any real movement on location data accuracy until the end of 2016 at the earliest.
The MMA is also calling for updates to the OpenRTB specs, a project first spawned under the auspices of the Interactive Advertising Bureau to create a common language enabling the buyers and sellers of programmatic media to more easily transact. In the intervening years, there have been a number of updates, the most recent of which, OpenRTB 2.3.1, was just approved in June.
The OpenRTB guidelines specify what information a publisher needs to pass along to buy-side platforms as part of a bid request. Although OpenRTB does, rather abstrusely, enumerate a few recommendations around how to transmit location data (i.e., ”The lat/lon. attributes should only be passed if they conform to the accuracy depicted in the type attribute”) there is no minimum threshold for what constitutes accuracy or precision and there is no mention of data freshness. The MMA is calling for both to be added to OpenRTB as required fields.
Data freshness is a particularly relevant parameter – it’s the difference between baseball fans, Sunday churchgoers and people walking past a Taco Bell. (Stay with me, it makes sense.)
First, take baseball fans. The 2015 World Series opener was more than five hours long. If you want to target someone at the game and you get a location read one hour in, it’s likely that the same signal will still be useful three hours later. Good to know.
Next up, people walking past Taco Bell: “If someone’s just passing by a store and the goal is to get them to come inside, a piece of location data may be completely irrelevant even a few minutes after you get it,” said Ubimo CEO Ran Ben-Yair.
And then, the churchgoers, because targeting isn’t the only location use case. There’s also segmentation and attribution. Mobile location data company Verve, for example, was once tasked by a movie theater to create an audience of people who regularly went to church on Sunday. That required observing users over time.
“You’re not sending churchgoers a message during a church service – that’s not going to happen,” said Verve CEO Nada Stirratt. “But knowing the device ID of someone who regularly goes to church on Sunday mornings means that we can create a segment in our database.”
In an ideal world, the location data necessary to power all three scenarios would be accurate, clearly labeled and priced accordingly.
That’s when the dollars will finally begin to flow, Lieberman said.
“Publishers, if you can prove to me that your location data is better than someone else’s, I will pay more for it – I will give you more money and more volume,” he said. “It’s checkbook diplomacy.”
Knowing Your Place
But before the buy side whips out its checkbook, there are a few more things to keep in mind – including knowing exactly what an advertiser is trying to get out of using location in the first place, said Hilary Maitland, director of client strategy at ThinkNear.
“Are they looking to drive someone into a store? Are they looking to drive someone down the path to purchase? What we recommend to clients most is to figure out what success is for them,” she said. “Understand contextually why you’re delivering an ad.”
There’s also a fair amount of education that needs to happen industrywide around how location data is generated. Despite the widespread notion that location data is really just a synonym for GPS – considered to be the best and most accurate location data point – it isn’t always easy to get and it isn’t always immediately 100% accurate.
Location is often comprised of a cocktail of data points. When an app is opened and location services, if enabled, are woken up, the app starts trying to figure out where a person is. The app pegs an approximation of location, and that gets more accurate over time. (It’s why when you first open up Google Maps, you see a large blue circle showing where you are that slowly becomes a tight blue dot. That’s the app triangulating data points – Wi-Fi, cell tower, GPS, etc. – to get a better read on exactly where you are.)
“The operating system’s location service gets better over time, and the longer a user uses the app, the better the data gets,” said Ubimo’s Ben-Yair. “Which is why this is not a black and white situation and why the best thing to do is to add transparency.”
But transparency is a two-way street. Publishers also need to be transparent with their users – if they don’t actually need location data, they shouldn’t ask for it at all.
“If an app has no reason to acquire location, users will understand that and they will turn [location services] off,” said Dan Silver, director of marketing at xAd. “Then it becomes useless for everyone.”
Speaking of the consumer, there was no discussion whatsoever of privacy, either in the MMA’s report or at its session – and and it’s something that has to be seriously considered, said Jason Kint, CEO of Digital Content Next.
While “selling location data to third parties should involve consent and be a thoughtful and informed decision” on the part of the publisher, said Kint, consumers also “need to know they’re carrying around a ubiquitous tracking device. At the same time, we’re living in the Wild West of ad tech tracking and targeting with significant access to data – moving further into the darkness with techniques like digital fingerprinting.”
