3 Common Audience Targeting Pitfalls and How to Avoid Them
Netflix created buzz in 2016 when the VP of Product Todd Yellin publicly described demographic data as nearly irrelevant. "Geography, age and gender? We put that on the garbage heap," he said.
While perhaps an extreme example, Yellin’s sentiment exposes the pitfalls of even traditional targeting methods.
Beyond the walled gardens of Facebook and Google, companies are creating audiences to capture mobile advertising dollars. How do brands ensure they aren’t buying a “garbage heap”?
Use Deterministic (vs. Probabilistic) Audiences
Pitfall 1: Many buy audiences built on probabilistic rather than deterministic data. Let’s compare the difference between the two:
Probabilistic data is data with a “high probability” of accurately categorizing behaviorally similar mobile devices into an audience. Generalizing high income zip code based devices as luxury shoppers is an example of an audience built on probabilistic data.
Alternatively, deterministic data offers a one-to-one match. For example, a device belongs to a luxury shopper because they’ve recently made several luxury purchases. While deterministic data can lack scale, it is inherently more accurate and valuable. With deterministic scale constraints, there is more inferred or probabilistic data within the mobile landscape.
Opt for audiences built from deterministic data when possible.
Avoid Audiences Built Exclusively from Location Data
Pitfall 2: Many buy audiences exclusively built from location data.
Location data offers a significant amount of insight about a mobile device ID and is extremely valuable to advertisers. However, its accuracy is convoluted, particularly in business-dense areas.
Let’s take a real-world example that exists in most major cities: Your brand wants to reach a health-conscious audience based on gym visits. There is fast food restaurant within 500 meters of the gym. With a location-only audience, your “healthy and active user” may actually be a “fast food diner.”
Ensure location is not the singular data point behind the audiences you buy.
Avoid App Library Based Audiences
Pitfall 3: Many buy audiences exclusively built from app library data.
A device ID’s app library lends an incredibly narrow data set from which to build an audience. For example, a Delta Airlines app on your mobile device does not automatically make you a business traveler. If fact, many infrequent leisure travelers often download an airline app to watch entertainment on their flight.
Buy audiences that are informed by a range of data points, not just app library.
Understand the Ingredients of the Audiences You Buy
Overall, when buying audiences, understand their ingredients.
Kiip’s built the audiences we would want to buy, thanks to our eight years of experience and our deep industry partnerships. Kiip Action Audiences are deterministic and leverage behavioral and intent data verified through audience surveys. Kiip Action Audiences are also:
- Mobile First - Built from 8 years and billions of mobile interactions across our network.
- Multidimensional Data Points - Unlike other audience vendors who rely on a single data point to create their audiences, we include data like timestamp, device characteristics, location, app usage, brand engagement, and survey responses.
- Verified - Kiip Action Audiences are built using signals generated by real people, on real devices, in real time. We use past engagement behavior and quick surveys to validate our segments and keep them fresh and accurate.
More from Kiip
Last week, Mary Meeker published her highly anticipated annual Internet Trends Report. It covers everything from internet usage, to e-commerce, economic growth, and consumer spending. If you don’t have time to digest all 294 slides, check out our key […]
Today we launched Kiip Action Audiences, the first-ever mobile audience product combining both real-life behaviors and survey-verified intent. Kiip Action Audiences are being rolled out through LiveRamp®, an Acxiom® company (NASDAQ: ACXM) and leading […]
Companies spent more than $20 billion on third-party data in 2017, a number expected to rise in 2018. There’s no question that we’re becoming more and more data-centric by the day, but are we properly evaluating this data? How can we actualize the […]