"Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Taylor West, general manager, North America, at Zeotap.
The eyes may be the mirror to the soul, but your smartphone is the mirror to everything else about you.
Given how tethered we are to our mobile devices, there is no better indicator of who we are. From what we install, to the frequency that we turn to them and the actions we take within, these precious insights form a picture of who we are as individuals. From interests, behaviors and even sociodemographics, our phones know us better than our closest friends.
Judging interests and preferences by app use is nothing new – it’s common to assume that people using the Food Network app enjoy cooking, and those invested in the ESPN app like sports. App use can also indicate sociodemographic information.
A potential wealth of insights
In 2016, researchers from Aalto University in Finland and Qatar Computing Research Institute analyzed demographics and app usage to create models to predict gender, age, marital status and income with up to 82% accuracy. For example, the study showed that women are likely to have Pinterest or Etsy on their phones, and if you use LinkedIn and Fitbit, your income is likely $50,000 or higher.
However interesting, these types of modeled insights – while a good starting point – aren’t nearly as impressive as what we could glean about audiences.
The industry has a tremendous opportunity to move beyond just a few data points about larger groups of the population to actually understanding the makeup of users. By looking at particular app usage and the frequency of actions taken, advertisers could be much more strategic in their planning by targeting the psychological makeup at an individual level.
What’s holding back in-app targeting?
The real challenge is making the most of app data for better targeting. Apps are closed, siloed ecosystems that don’t speak to each other, so advertisers don’t always have access to a broad profile across any given user. For example, through SDKs and other means, it’s relatively easy to tell whether Pinterest, LinkedIn and Uber are installed on a user’s phone. What’s much more challenging and out of reach is whether they actually use these apps and how.
What if the user hasn’t opened 40 of their apps this year? What exactly are they doing in the apps they do use? In general, SDKs have limited insights into actual use of apps, based on the permissions granted to them from app creators.
The other issue is that popular premium apps such as Uber are very protective of their data, but it could be a game-changer for advertisers to know users’ travel behaviors. For example, if a brand wants to engage with a younger crowd that frequents clubs or bars, it might target users between 21-34 years of age who have higher propensities to use the Uber app frequently during late evenings or early mornings on Fridays and Saturdays in urban areas.
If we ever want to truly tap into the targeting gold mine of apps, the industry needs to solve the technological challenges of cross-app silos and convince premium apps that their valuable data can be an incredible revenue stream while still remaining privacy-compliant.
Advertisers say they want one-to-one relationships with their audiences and consumers say they want more relevant ads. We simply need more visibility to piece app data together to make substantially more insightful user profiles.
Everyone has data privacy and the upcoming General Data Protection Regulation on their minds – as they should – but we must find the right balance to dramatically improve digital advertising from modeled assumptions to those based on how users actually think and act.