Today’s ads allow two-way conversations where consumers provide direct feedback to a brand. Ad creative has evolved significantly from the MadMen days of Don Draper, when advertisers only made outbound statements to consumers. Active preference ads are one example of an interactive ad that exists today. These ads ask viewers to – through a direct declaration – describe their current preference with respect to a specific brand, product, or offer. Each ad poses a multiple-choice question; when the consumer responds, the brand gains a real-time and directly-relevant understanding of his or her preference.
Think of an electric toothbrush company that releases an electric flosser product. In order to qualify the audience and avoid wasting impressions to those who are not intenders for a flossing product – the brand must distinguish between those who floss their teeth versus those who don’t. An active preference ad would allow consumers to sort themselves by answering this question.
These ads also give consumers the opportunity to voice how often they floss as well as whether they have frustration with regular dental floss. These two options help a brand to determine the likelihood of the consumer purchasing an electric flosser, and to disregard or target him or her accordingly.
Active preference ads drive performance, deliver insights, and – of the four types of positive response that can be modeled (below) – are the most accurate and the best predictors of response. Here’s why.
Third party data is the most readily available, but, as a stand-alone solution, it is the most challenging to use. Third party data providers build their segments to be generally applicable to many marketing efforts and therefore include as many members as possible. Each segment paints a broad and all-inclusive picture of its members. There may be some cases where an advertiser’s product aligns well with a prebuilt third party segment (think of Maserati targeting its car ad to the IAB’s “Automotive: Maserati” segment), but most segments do not align perfectly with the product, service, or offer being advertised.
Consider a brand that manufactures designer watches. Because these watches are more expensive than others on the market, the brand chooses to advertise to certain income segments. Within those segments, though, the brand has no way of knowing which consumers are in the market for a new watch. Or whether the consumers even wear a watch in the first place. An active preference ad would allow the brand to qualify consumers according to their interests, disregarding consumers who don’t need a watch as well as those who are satisfied with their watches now, and focusing on those consumers who are in the market for a new watch. By running an active preference ad prior to the start of the brand’s campaign, the advertiser might learn that, within a certain income segment, the the true intender persona skews toward a male in his thirties who is in a professional business role and interested in luxury.
Site pixels provide some information about visitors who have shown interest in the specific product or service, which is an improvement from segment targeting. Site pixels usually provide a strong enough signal to build a predictive model, and they allow an advertiser to target more people like the site visitors. However, they give no indication of whether the profiled site visitors are the best audience for ads. Let’s say that a department store runs ads targeted to fathers for Mother’s Day and to mother’s for Father’s Day; it deduces that the other parent is responsible for purchasing presents for each holiday, and the brand is right – click-through rates soar. Both ads perform very well. Later in the year, the brand runs a Christmas campaign to consumers who have visited its website throughout the year. The creative shows an ad for male and female sweaters. It performs poorly. This could be for any number of reasons – parents are shopping in-person; parents are more focused on buying presents for their children; parents are buying jewelry and home improvement products more so than clothing; parents only make up a small portion of website visitors throughout the year, and the other consumers aren’t interested. Regardless of the reason, the brand isn’t able to determine preference by simply having a site pixel on its site.
Campaign response is the next best type of data. By combining third party data with both positive and negative campaign responses, machine-learning systems can identify intersections of several segments that truly differentiate responders from non-responders. These models can be highly effective in predicting response for targeting. Campaign response shares one weakness with site pixels, though. While it can differentiate responders from non-responders, it has no visibility into why consumers act the way they do. Response may be muddled by attraction to the creative or boredom with the surrounding content. Ad response does not have any information about the consumer’s actual and immediate preference for the offered product over a competitive alternative. Let’s say that an online travel agency wants to market all-inclusive Caribbean packages. In order to reach intenders, the brand shows ads depicting a couple on a beautiful beach. Responders will include people who like pictures of beaches and people who aspire to travel to the Caribbean someday. While they are responders, not all of them will be qualified prospects.
Active Preference allows consumers to respond to a differentiating question to show the option that they prefer. A kitchenware retail brand, for instance, might have a new waffle maker that it wants to advertise. It could start by determining the active preference of all consumers with a question like the one below.
Responses to the question in in the creative clearly separate those who prefer pancakes from those who prefer waffles, as well as those who are potential purchasers of an at-home maker versus those who tend to dine out for breakfast. Once the advertiser collects a set of people who prefer to make waffles at home, it is reasonably straightforward to correlate those active preferences with first and third party data and build a model that is very good at identifying the consumers who they should target with a waffle-maker ad. Ultimately, this is the most effective way to drive responses that continue to conversions that lead to bottom line revenue.