This article originally appeared in MarTechSeries.
Unstructured data, simply stated in advertising terms, is any piece of information or behavioral signal collected in its simplest form. Once an assumption is made about the person behind that data signal and it coalesces with other signals, it then becomes segmented data or a “Data Segment.” This is a form of institutionalized stereotyping common to advertising that was born out of a technical need to generate scale and speed. Data providers, including Google and Facebook, commonly curate vague taxonomies like “auto intenders in Austin, Texas” or “high-income retail enthusiasts.”
Perhaps the recent Cambridge Analytica scandal prompted you to download your Facebook data. Did you recognize yourself? When you take a large number of signals and group them together, you are going to get some stuff right and some stuff wrong. Segments are built on assumptions, and they can be wildly inaccurate. For marketers, that means you are wasting impressions by serving targeted ads to people who don’t actually match your criteria.
It’s true that audience segments can make workflow faster and scale more easily achieved. These advantages come with an all too common consequence repeated by digital advertising companies campaign after campaign. “We can hit your goals provided you spend enough or let the campaign run longer.”
Unstructured Data Unlocks Important Capabilities on Mobile
The rapid growth of mobile internet access via smartphones has unlocked a significant trove of new unstructured data signals. The most notable is actual GPS data verified by at least 3 of 31 GPS satellites or verified via beacons and more. According to GPS.Gov, typical GPS-enabled smartphones are accurate to within 4.9 Meters or 16 feet under an open sky. So, mobile location data can be very accurate and there is a lot of it!
Research from comScore found that mobile accounts for 69 percent of total digital media time spent, whereas desktop has fallen to less than one-third of digital media time. For advertisers, mobile is a whole new world. Mobile location data tells you where someone is and/or where they have been, which can be like a blinking red light of intent.
But many marketers today are knowingly or unknowingly executing campaigns utilizing mobile location data curated into bulk “audience segments.” This approach is in direct conflict with the potential of accurate location today to produce better campaign performance faster and with fewer impressions. In fact, a recent study found that marketers’ biggest mobile advertising challenge is opaque data: a lack of visibility into the data that was used to define their audience targeting. The fact is that while the location data can be very accurate, most mobile ad networks’ access to data is limited to small subset of mobile users. In the spirit of scale, the granularity of the data gets pooled together into a larger than desired targeting area, typically with users visiting that area over an extended period of time. Hardly precise.
Here’s what you are missing if you don’t have a means for tapping unstructured data.
The Size of Targeting Areas: Geo-Fencing for the Real World
Marketers use geo-fencing to target mobile users within a particular area. The average square footage of a quick service restaurant is between 2,000 – 4,000 square feet and rarely is the rooftop or plotline a perfect square or circle. You know your approach is segmented data if the solution being deployed required a fixed radius that is much larger than the business you intend to target. In fact, many location data companies use fixed-shape geo-fencing products that are some form of grid-based data storage and retrieval.
The world’s most precise location data isn’t very precise if it’s deployed at a size and shape that isn’t what is desired or intended. Geo-fences that use unstructured data make it possible to create custom shapes that are specific to the unique area you desire to target regardless of size. This results in more accurate targeting and fewer wasted impressions.
The Importance of Data Recency: When Matters as Much as Where
When you are targeting mobile users, there is a big difference between a prospect who was at a location a few hours ago and one who was there 27 days ago. Recency matters! Location data companies collect time-stamped data from mobile devices. When it is unstructured, marketers can customize their messaging and their bidding strategies based on the recency of when that consumer was in the targeted area. When it is structured, all that data is lumped into the same black box that forces the optimization to treat data that is three minutes old the same as data that is 60 days old.
Let’s say a car dealer wants to create a geo-fence around a competing lot. Someone who was at the location four hours ago is more likely to still be in-market for a new car than someone who visited the lot three weeks ago, so you may want to bid more aggressively for the more recent impression or perhaps serve a different creative message. Factoring recency into your strategy can deliver more cost-effective results than just targeting a broad segment of “auto intenders” who have visited nearby lots within the last 30, 60 or 90 days.
The Cross-Device Promise
When users are on a browser, behavioral data is collected and stored against an ID, and we begin to form a pretty robust understanding of how people are spending their time online. Each mobile device also has an ID, and cross-device matching allows ad tech companies to map the behavioral data they’ve been collecting from browsers to the appropriate mobile ID. This is an oversimplified explanation, but the point is that the potential of cross-device matching allows for an injection of even more unstructured data signals to be used to refine your location-based marketing campaigns.
Let’s say you are a national financial services company with independent financial advisors all over the country. If data is unstructured, you could serve mobile in-app ads to users who are in the vicinity of your local offices and have searched for “financial planning” in the past two weeks.
Marketers value mobile and location data because it’s precise, but you lose so much granularity when you structure it. Using audience segments is the legacy way of approaching audience targeting, but it is time to evolve. Marketers need the ability to execute on the realities of the real world and to refine their strategies on mobile the same way they do in search with keyword data. Without unstructured data, you cannot realize the full promise of mobile.