Consumers in this fast-moving world are constantly changing their desires and habits. Today, you may be in the market for soccer shoes. Tomorrow, you’ll have moved on to bathroom hooks. Such moving targets are both a challenge and an opportunity for marketers.

What are Look-alikes?

Look-alike audiences, a targeting approach that allows modeling a large audience based on a smaller one with similar attributes, has become the bedrock targeting practice for programmatic advertising, as well as several social media marketing channels. Look-alike modeling is an excellent tool that allows you to use a set of owned audience information to prospect new customers from a third-party data set.

But basic look-alike audiences are only the beginning of what can be accomplished when it comes to modeling. I believe that a more creative, custom approach to audience modeling is the future of data-driven marketing. Here’s why.

Basic look-alike modeling is achieved by creating a “seed set” of your own customers, then using third-party data from a data provider or a marketplace to build larger audiences with similar attributes to your seed set. The problem is that many are doing look-alike modeling. Others are buying the same off-the-shelf audience segments from the same DSPs [demand-side platforms for buying online ads] and data providers, or are using black box solutions that offer little transparency into their look-alike algorithms. This means the exact same data is informing your strategy—and these you sell against.

The Custom Difference

By contrast, a custom approach audience modeling is a more highly crafted method of building a high-value audience with finer segmentation, more strategic data partnerships, and individually optimized creative. Here are some key custom audience modeling strategies:

When it comes to defining a “seed set,” many marketers miss the opportunity to take a strategic approach to data segmentation. While it’s easy to model an audience off all of your customers—or perhaps all customers who’ve spent over a certain amount—taking a more nuanced approach to segmenting your customers can allow you to target your audiences with greater relevance.

For example, a men’s apparel brand may have drastically different male segments they keep in the same demographic data bucket, such as “stylish buyers” who will pay a premium for the latest fashions, “deal-seekers” who buy items only when they’re discounted at end of season and “gifters” who spend a fair amount but only during holidays. Yet, each of these segments could be that basis of their own carefully refined seed sets The creative and messaging crafted for each segment may be completely different in order to better meet the needs of each group.

The Importance of Partnership

In order to implement a custom approach, marketers should seek out a third-party provider whose data can provide the greatest incremental value to their own first-party data. In some cases standard data sets and segments available on open data marketplaces can be quite useful (and may be more cost-effective); however these are often used by many other marketers, meaning your audience could already be receiving a lot of advertising from competitors.

Consider a strategic data partner, such as providers who specialize in a particularindustry, or who have extensive experience in the kind of campaigns and KPIs that matter most to you. For example, if a retailer is interested in increasing conversions on a particular product, they may work with a partner who has “declared” purchase data—audiences known to have actually bought this product before.

A strong data partner can also help you better understand and enhance your own first-party data by layering their larger audience set on your own to enrich your own customer profiles.

The Last Touch

The final, crucial aspect of custom audience modeling is having the talent—and technology—to optimize in real time to constantly improve the targeting of your audience. Ad creative should be dynamically served and fine-tuned as the campaign progresses, as members of the audience are scored, and subsequently added or subtracted as new data points flow in. A stream of fresh profiles are delivered as they emerge, and profiles that are no longer relevant are removed in real-time.

“It’s not just about prediction. It’s about refinement and calibration,” says Mark Donatelli, Global Head of Data Strategy & Planning at Ogilvy & Mather. “Targeting an audience is similar to a NASA moonshot, NASA may calculate the course of the rocket, but it needs to keep adjusting the heading to stay on course. Whether it’s a moonshot or a marketing campaign, You can’t just pick a point and go.”

Connexity is a digital consumer insights and activation platform designed to help marketers find, get, and keep customers.

Favorite