AVON is a 131 year old American international direct selling beauty company.
-80% overall CPC reduction
700,580$ Estimated budget reclamation
+272% increase in ROAS
AVON North America is comprised of two core business functions, the recruitment of in-field AVON representatives and a sizeable e-commerce business via AVON.COM. The overall objective was to reduce the cost of Brand by testing moving from dual auction entry to a single auction entry reducing CPCs for AVON’s two business functions. It was critical that overall Revenue, Lead and Contract volume driven through branded search in a dual auction setting was maintained.
Before any planning was possible we structured a cost saving and benefit’s hypothesis with defined benchmarks and success indicators. Product downtime resulting in media pauses allowed us to extrapolate expected CPC and QS impact.The hypothesis was proposed and approved for testing against a rigorous testing schedule.
Due to internal politics between AVON and AVON’s most successful representatives, the GWP solution was not approved to show for previous AVON.com customers that were shopping attached to a representative. In order to ensure that only unattached customers received the single auction entry, we used in-engine data functionality such as customer match to negative the attached customer audience. Additionally, and in partnership with internal technical teams, we started to pass real time audience updates from DoubleClick through a piggybacked AdWords pixel to an AdWords remarketing audience. These updates were based on user status on check-out ‘event’ to ensure data integrity and ‘freshness’ was maintained. The attached customer audience was then targeted through separate brand campaigns with the old dual entry audience experience.
Due to the complexity of the testing, it was not possible to use AdWords experiments in order to AB test the different experiences. As such, we set out a testing schedule running the GWP experience for 25%, 50%, and 100% of daily search impressions at different times of day and different times of week over a period of 4 months. The extensive testing schedule was structured to negate the impact of external variables ensuring that Revenue, Lead and Contract volume was consistent with the dual auction entry benchmarks.
Through collaboration with internal creative teams, multiple test iterations of the page were developed in an effort to not only prove the hypothesis but improve overall performance through LP optimisation. This resulted in more simplistic versus more information heavy pages being tested sequentially to a share of traffic over the course of a number of weeks.
Against original parameters, success was tracked in real time to confirm impact on product performance. Using Doubleclick Campaign Manager, out bound clicks were tracked as separate campaigns in order to ensure that costs (now shared) could be apportioned correctly. Additionally, DCM Floodlight data allowed for the tracking of leaked events via static page elements.