The Future of Targeting Relies on the Right Data
For the past decade, ad targeting has remained relatively unchanged. In its most rudimentary form, a marketer uploads a first-party segment list in the form of a CSV file to Google or Facebook and targets that segment with the creative that the marketer believes will convert the most with that audience. Ultimately, the decision of who gets shown what ad is predetermined by the marketer.
However, ad targeting is about to undergo a seismic shift. In the not so distant future, marketers won’t select who is targeted. Rather, machines — powered by ML/AI — will determine the ads consumers see.
Marketers have always created custom data segments for campaigns (e.g. women under the age of 35) and have selected the content that segment is shown. But today, everywhere we look, the decision of whom to show a piece of content to — which in the ad world is pre-determined by marketers — is increasingly being augmented or replaced with machine learning to optimize who gets shown what ad.
Those who use digitally native platforms like Netflix or Spotify or TikTok are already at least somewhat familiar with how this works. The first thing a user sees when they open up TikTok is the platform’s “For You” page, a personalized stream of videos, with no two For You pages being exactly alike. TikTok says it relies on a complex set of weighted “signals” to inform its recommendation system, such as interactions with videos you like or share, accounts you follow, comments you post, and content you create.
Before we get to that future of targeting, there will be a natural next progression where the data that we use to inform segments will be augmented or enriched with custom data, such as signals, to improve the creation of those segments.
Think Personalization, not Segmentation
Today, marketers use first-party data from a number of different sources such as their website, mobile app, CRM, e-commerce platform, and point-of-sale (POS) to inform segments. We call these smart segments, and they surface data points that turn traditional segmentation upside down. This data is being used by marketers more and more to enrich segments and to recognize affinities and make predictions about what a prospect is likely to do next.
You are probably familiar with the concept of affinity audiences as targets with similar interests, like people who drink coffee. What’s different about these affinity audiences is the type of data that’s being used and the qualities we are looking for in that data such as whether a person has an open customer support ticket, has made an in-store purchase over the past two years, purchased $500 worth of products, or visited the website over the past 30 days.
Once marketers have all this information, they can build custom segments based on behavioral insights that they have gleaned from this data, make predictions on how that segment will behave, and identify hidden trends and patterns that they never thought of before. For example, a retailer could know exactly who is showing interest in a particular sneaker, where they are located, whether they’ve purchased in the past year, and level of interest in a product by merging data from all of these platforms into a single view.
However, in order for marketers to have these insights, and make discoveries about customer data to inform these segments, they need to be able to access and explore it. Often, this work is done by an analyst and the marketer doesn’t have access to the work. This absent workflow means activations often lack insights and marketers are not able to connect their data with their audiences, and their targeting suffers.
Collect the Right Data
Collecting this data, and having a process for using it in your marketing, is key as we move towards the next phase of targeting, where the machines will dynamically select the right ad to serve the right consumer. In order to get there, marketers must make sure that they are collecting the right data today.
Traditional customer segmentation often suffers from poor data quality and management that can result in wrong decisions and wasted marketing budgets. Using AI to do the targeting, marketers will generate better results and can eliminate human bias when analyzing data to create segments.
James McDermott is the co-founder and CEO of Lytics.