After running a three-month test with Dynamic Yield for sitewide recommendations, the Build with Ferguson team wanted to define and implement a clear targeting strategy for personalized recommendations. They did this by adopting Dynamic Yield’s Root Audiences, an internally-developed macro-segmentation framework designed to bring personalization to life in a functional, repeatable, and scalable manner.
The team believed different audience segments have varying, distinct needs when shopping online and therefore projected implementing a singular recommendation strategy for all users site-wide would be ineffective. After adopting the Root Audiences framework, which required the team to identify macro audience segments to target in its recommendation campaigns, this theory proved accurate. It helped their team identify distinct behavioral differences between segments and glean insights and learnings the team could then use to optimize its recommendations.
For example, Root Audiences helped Build with Ferguson identify different behaviors across their two core users: Trade Professionals and Consumers. Trade Professionals, who turn to the retailer to purchase everything they need for an entire home, building, or construction site project, engage the most on average and account for just 1% of total site visitors. On the other hand, Consumers represent the average Build with Ferguson site visitor and often are in search of particular products for specific projects. Before adopting the framework, the team treated all users the same when it came to product detail page (PDP) recommendations. However, after taking this macro-segmentation approach, the Build with Ferguson team was ready to test serving different experiences to each type of core audience to optimize these digital experiences.
The Build with Ferguson team began to test various recommendation algorithms and experiences for both Trade Professionals and Consumers across the site. For example, they served PDP recommendations using the ‘Recently Viewed’ and ‘Viewed with Recently Viewed’ algorithms to Trade Professionals and a different set of recommendation experiences using various algorithms including, ‘Recently Viewed,’ ‘Affinity,’ and ‘Viewed with Recently Viewed,’ to Consumers based on users’ levels of engagement.
Several interesting learnings emerged. First, the team noticed Trade Professionals tend to engage with recently viewed products on the homepage to navigate back to PDPs they recently interacted with. Second, the average user (members of the Consumers segment) tends to engage with recommendations based on items other users with similar behaviors and interests have interacted with.
Using these findings, the team optimized the performance of their recommendations across the site. By delivering more relevant, engaging experiences, Build with Ferguson was able to consistently connect all customers to products that fit their needs, resulting in an 89% uplift in purchases generated from recommendations.
Build with Ferguson extracted additional insights, most notably that users who interact with recommendations spend 13% more and purchase 2.4 more items on average. This data will inform how the team designs future campaign variations and tests.