What Makes Better Behavioral Targeting Segments?
Simpli.fi CEO Frost Prioleau says his company engineers have developed a method to improve the way advertisers buy behavioral segments.
Prioleau calls it de-averaging behavioral targeting, a process that, theoretically, allows the audience segments to generate better results with a narrower focus. He explains the process as "the ability to understand the performance and value of micro-segments within an overall BT segment, and to vary bid prices based on the performance of each micro-segment." Optimizing on high-performing micro-segments, and eliminating or lowering bids for the low-performing micro-segments, can greatly improve overall campaign performance.
For instance, say the Waldorf Astoria runs an online behaviorally targeted campaign. It may procure a typical behavioral targeting segment for a "New York hotel intender." Data from many travel sites would typically comprise this segment, and would include users who have expressed interest in NY hotels ranging from a Holiday Inn Express to Hilton to the Ritz Carlton, Prioleau explains.
"If they are able to de-average this campaign, they may very well find that users who expressed interest in the Ritz Carlton were much more likely to convert than users who expressed interest in a Holiday Inn Express," he said. "In that case they can improve campaign performance by eliminating or lowering bids on Holiday Inn Express users."
This type of optimization can improve campaigns by at least 100% because a client may load 1,000 or 10,000 keywords into the system, and then see how the campaign performs for each keyword on a CTRs, CPCs, and CPAs. Each keyword becomes a micro-segment, representing users who searched on that particular keyword.
Prioleau believes better performance for BT requires a platform that allows:
1) Advertisers to quickly build and launch custom BT segments.
2) Transparency on how each BT segment performs including spend, clicks, and conversions.
3) Instant editing of BT segment without the need to rebuild a cookie pool each time a component is added or removed.
Priolea also suggests some best practices for adverisers to follow when implementing BT:
1. Begin campaigns with broad data and move to narrow based on results seen in the campaign's initial stages, but remember a very narrow data set can prevent it from reaching its full potential. Effectiveness of various data varies with different creative pieces and offers.
2. Use site data to model lookalike segments when possible to enhance BT campaigns. Often effective data for BT campaigns is gained by finding users who have similar attributes to users who have already clicked or converted on an offer.
3. Don't forget the impact of where the ads show. Remember that ad position often plays an important role in campaign performance. Once the campaign runs, use site-level performance data to eliminate poorly performing sites.