Many marketers are seasoned in retargeting and direct response marketing challenges. Scale and accuracy tradeoffs, and a predisposition to bot fraud, can make retargeting a difficult strategy.
Even ShopStyle, a leading online retailer with notorious digital marketing chops, has confronted these problems. ShopStyle, a division of PopSugar, is a popular shopping search engine and sales aggregator that drove $1 billion in revenue to its retail partners last year.
Earlier this year, ShopStyle approached TruSignal and MediaMath for a solution to improve direct response and retargeting performance. ShopStyle already had campaigns in place on MediaMath’s TerminalOne Marketing OS™. With MediaMath’s and TruSignal’s powerful integration, TruSignal leveraged its offline data, predictive scoring and people-based audiences to improve ShopStyle’s retargeting campaigns:
- Focusing on converters—Retargeting campaigns advertise to consumers who previously visited a website. ShopStyle had the daunting task of sussing out which of its 18 million monthly unique visitors are good fits for the brand and which are merely window shoppers. Combining predictive scoring using offline data and the online behavioral signals delivered a more comprehensive retargeting strategy, focused on finding visitors who aligned to the brand and its overall marketing objectives.
- Reducing fraud—According to the Association of National Advertisers, fraudulent traffic could represent up to 37 percent of those impressions. People-based audiences are linked to verified consumer profiles, which allows retargeting campaigns to hone in on real people, not bots.
- Increasing bidding accuracy—Many retargeting campaigns assume every visitor hitting the same web page is equally valuable. Most marketers have no way to know which ones are more likely to convert than others. So bidding on these impressions based on actual worth is extremely difficult. Using offline data to create a predictive score for purchase propensity boosts a DMP’s or DSP’s ability to value a bid based on individual consumer value.
In order to suss out which visitors were better brand fits, and to create a solution unique to ShopStyle, TruSignal started with a sample of the retailer’s best existing customers.
TruSignal used offline consumer profile data and its predictive scoring engine to analyze thousands of data attributes per customer, weighing them according to relative importance, in order to build a custom predictive model. The offline data provided a comprehensive view of known consumers that complemented the online behavioral signals and delivered the massive scale of people-based marketing.
TruSignal used the predictive model to calculate a zero to 99 score for 220 million U.S. adults, based on likelihood to convert for ShopStyle. Using the lowest scores, zeros through 50s, TruSignal created an audience filter representing people unlikely to convert. With the help of MediaMath, ShopStyle applied the audience filter as a negative targeting condition to ShopStyle’s existing retargeting campaigns.
By filtering out non-converters, ShopStyle saw an increase in conversions of 200 percent. In addition, ShopStyle reduced the Cost Per Acquisition by 60 percent. MediaMath integrated the custom predictive scores into the retailer’s bidding algorithms to improve the overall bidding strategy. Each predictive score uniquely influenced the price paid for each impression, which drove more accurate bids.
The audience filter’s effect was tested using a side-by-side test and control methodology. ShopStyle created two identical groups for comparison: one of their existing campaign without the filter, and one of the existing campaign with the applied filter. The groups were kept mutually exclusive to ensure the results showed conversion difference that was solely attributable to the use of the audience filter.
Download the full case study here.