Modeling Behavior To Target Smarter
Most online marketers now embrace behavioral-based targeting. Not long ago, only a handful of technologies existed that enabled marketers to align their online targeting strategies based on a person’s browsing behavior. Today, there are over 200 online vendors that drop cookies, a tiny piece of unnoticeable code on browsers. Using signals passed from this code, vendors recreate a person’s browsing behavior.
This type of behavioral tracking became extremely appealing for marketers, who could now target ads based on which type of site a person visited, time of day, and how often. As more and more marketers focused their online budgets to target based on behavior, the price for specific behaviors became more and more expensive. Automakers began paying top dollar for any user who visited an automotive site; they paid even higher premiums if users visited multiple sites on several occasions. Yogurt producers may pay top dollar for users who frequent health sites seeking fiber advice, etc. The truth is that a lot of these tracked users are well underway to engage with the brand and are likely to convert anyway-- in short, they’ll buy no matter what. If that’s the case, then marketers focused exclusively on behavioral targeting may be spending their time and money in vain.
Behavioral targeting no longer can work in silos as the core targeting strategy. Targeting is more complex now, the consumer is savvier online, and a lot more data is made available. I suggest considering a three-tiered approach to targeting:
Tier1: brand building to create the demand and drive momentum,
Tier 2: behavioral modeling to reinforce your message, and
Tier 3: predictive models to better understand which consumers will likely convert without any promotional help or media disruption.
Sequencing your targeting approach will be important, so I recommend the following:
1) Quantified brand building – We all know brand-centric media provides a halo benefit in supporting targeted media. However, I would argue that you can also use brand-centric media for targeting. One of the best ways to do this: get in front of a consumer’s decision cycle. Determine that critical moment when your brand, message, and offer makes the biggest impression to influence your consumer’s decision. To do this, you need to evaluate historical log files and define the exposure window, from the first time your ad was delivered to the conversion (or success event). Exposure windows will vary by campaign and by product. It may be 60 days for big-screen TV or 90 days for a financial product. You can use these exposure windows to get your brand message out there in a strong way early to instill, influence and maximize the halo benefit.
2) Behavioral modeling – Once your media is in-market for a few weeks, your campaign starts to generate a data trail. Behavioral modeling rebuilds the path to conversion by stitching together multiple campaign touch points. There are several versions of behavioral models, most of which group common online pathways into consumer profiles. For example, profiles can differentiate a male prospect in-market for a luxury vehicle versus a female shopper searching for yogurt. This sort of modeling enables media to follow the consumer based on revealed patterns and promote the brand message, or reinforce a more focused message for the subset of consumers who have already visited a brand’s website and showed interest.
3) Predictive modeling – One of the worst media tactics is to present a discount and/or offer to consumers well on their way to converting already. Instead you would want to thank your likely converters for engaging with the brand. Using predictive models make the brand smarter by understanding consumers' needs and next steps. Predictive models could be developed on the very same data you already store and use for analyzing your exposure windows and developing behavioral models. Whether you store the log level data in-house or data is stored on your behalf, getting access to it for modeling should be simple. Having a predictive view into your consumers' performance will inevitably change your targeting strategy. For example, you may test a more aggressive offer to a consumer less likely to respond, or customize the experience for a consumer more likely to convert and not focus on discounts/offers. Finally, for a consumer you know is a loyalist and will buy, you may simply choose to save your media dollars and stay out of the way.
Using data to help define a three-tiered targeting approach will enable you to better understand your consumer’s online journey and be more customized and efficient in delivering your media.