From Retargeting to Pretargeting – What is Predictive Advertising?
Rapid advancements in online advertising technology have given rise to a number of sophisticated targeting methods that have proven to be effective in reaching users based on their patterns of online behavior. Advertisers can now serve ads to users who have visited their company’s website or performed a product-related keyword search; this practice is known as retargeting, and has gained significant traction due to its ability to improve advertiser ROI. Although retargeting has proven to be effective, it has its drawbacks in that it keeps the advertiser in a reactive posture due to only being able to target user actions that have already taken place. But how can an advertiser reach all of those potential customers who haven’t yet revealed their buying intentions?
The answer lies in predictive advertising. Also known as “inference advertising” or “anticipation marketing,” predictive advertising enables brands to make advertising decisions based on not just one particular piece of data (i.e., a user visiting the company website), but on a wide range of information comprised of several different attributes and factors. This robust data science offers actionable predictive analytics that can help advertisers accomplish the coveted objective of serving the right ads to the right audience, at just the right time. Through advanced data analysis, companies can now build a profile of their target audience that includes key insights based on behavioral, environmental, socio-economic and psychographic information. This data informs the advertising process, resulting in increased conversions and decreased costs.
Target: Predictive Advertising to Expectant Mothers
As an example, Target has utilized its data-crunching prowess to scrutinize sales patterns and formulate a predictive model that can tell them when a female customer may be pregnant, long before she starts buying diapers. Target discovered that women who are in the early stages of pregnancy typically purchase a combination of 25 different products that include cotton balls, unscented lotion, and vitamins. Buyers who exhibit this purchase behavior are then sent coupon booklets via mail for items that are typically needed later in the pregnancy, such as nursery furniture and maternity clothes.
Nike’s Predictive Advertising Case Study
Nike partnered with mobile advertising specialists UberMedia to develop a campaign promoting NBA star Carmelo Anthony’s new shoe. Users who followed the New York Knicks on Twitter (both the team account and individual players’ accounts) were targeted, as well as smartphone users whose phones were tracked via GPS on tennis courts, running trails or hiking trails within the last 30 days. In addition, whenever smartphone users in Manhattan were within a certain proximity of either a Nike or Foot Locker store, an ad would appear on their phone to direct them to the most relevant purchase location.
UberMedia: The ‘Big Brother’ of Advertising?
UberMedia’s service utilizes geo-location info and social media site data from smartphone users to build a trackable profile of preferences and behavioral patterns. Ads are served intermittently throughout the day, automatically adjusting to fit the most relevant data points in the user’s profile. While many critics point out the borderline-Orwellian implications of such granular data tracking technologies, UberMedia insists that it is using its powers only for good. Bill Gross, the CEO of UberMedia, has issued several statements to address the well-founded privacy concerns that have emerged. Gross emphasizes the fact that the service is opt-in only, and any users who prefer not to be tracked can turn off UberMedia’s default tracking functionality in their phone settings, or just send UberMedia an email requesting to be removed from the list. Gross also points out that tracking is completely anonymous, and no user information is sold to any third parties.
Predictive Advertising: The Next Step
So what’s next for predictive advertising? Savvy brands are beginning to use their retargeting data to build model user profiles based on proven historical data; this is known as “look-a-like” targeting, and is frequently utilized in predictive advertising. Many industry analysts claim that the science of machine learning will help to automate the optimization process for predictive advertising campaigns. Right now, manual optimization is more the exception than the rule for most advertising campaigns, but with further technological development, the day may come where the entire predictive advertising process (including pricing, targeting, etc.) will be 100% machine-driven, and able to be scaled across several millions of impressions within seconds. As data science continues to advance, predictive advertising will continue to become more prominent, and perhaps eventually indispensable to the online marketer.