SUCCESS STORY

AppNexus Programmable Bidder

Turbo-charge incremental customer lifetime value

The Challenge

Numberly, a programmatic CRM company helping brands collect and activate first-party data in order to maximize advertising ROI , worked with a global health insurance leader to optimize the customer lifetime value (CLV) of its high-value customer base.

Numberly’s CLV Maximizer® product helps brands contact and increase the lifetime value of existing customers with a combination of programmatic media and email. With CLV Maximizer®, targeted anonymous customer segments are onboarded. A test group is exposed to multichannel, multi-format engagement across devices – allowing for measurement of incremental revenue generation, in comparison to a non-exposed control group.

Because successful CLV lift requires surgically-precise targeting of closed populations with a pre-defined budget , a laser-focus on locating and engaging target users is required.

For example, if a user has only been exposed once, a higher bid is set to re-enforce the message and to increase probability of engagement.

In a standard AppNexus campaign, new campaigns exclusively target segments containing users identified as having only been exposed to ads once. Once a user has been reached, a new campaign is set up to target users who have only been exposed twice. This approach requires substantial manual setup and demands additional management from ad operations teams.

AppNexus logo

The Solution

Numberly engaged the AppNexus Programmable Bidder (APB) and its flexible machine learning infrastructure to build optimization models that incorporate recency of ad exposure into bidding decision logic.

Leveraging a new recency parameter within the Bonsai language used to build decision tree models, Numberly uses the number of minutes since a user’s last exposure to calculate the appropriate bid for an impression opportunity.

Dramatically simplifying KPI achievement, Numberly builds decision trees consisting of several branches, with each one corresponding to an interval of recency . A specific bid is set for cookies exposed for less than one hour ago, while higher bids are set for cookies exposed in preceding incremental recency periods. This tree design solves for the n-th exposure, ensuring a successively higher bid until it wins the auction, thereby decreasing the risk of missing a high-value user over an extended period of time.

With APB, Numberly can manage all exposure – recency permutations for its advertiser within a single campaign. This results in massive time savings for ad operations teams, which can be re-invested in campaigns leveraging Numberly’s proprietary incrementality models.

The Results

To test the efficacy of the new recency parameter available in APB, Numberly executed an A/B test pitting campaign A – consisting of bids without recency logic—against campaign B – consisting of bids informed by the recency parameter.

Using APB, Numberly’s client experienced dramatic performance gains in both click-through rate and conversion rate metrics .

The campaign using APB recency trees experienced a conversion rate 65% higher than the reference standard campaign benchmark.

The APB campaign also realized a 63% increase in click-through rate performance, substantially exceeding advertiser expectations.

Key Figures

63 %

increase in click-through rate performance

65 % higher

Compared to the reference standard campaign benchmark, the conversion rate was