How LXRInsights and the Big Dee’s team turned customer intelligence into a biweekly experimentation habit, and a steady stream of measurable results.
OMMA Awards 2026 FinalistRecognized for AI-powered experimentation →
Same passion. Bigger possibilities.
Data as of October 2026, continuously updated every 30 days
Some regions were down significantly, with generic targeting missing key buyers.
Broad targeting diluted spend instead of finding high-intent customers.
Lapsed and at-risk customers represented an untapped revenue opportunity.
Combined Big Dee’s first-party data with LXRInsights predictive intelligence.
Focused spend on high-intent buyers for proven products in key markets.
Used predictive models to identify and re-engage high-value, at-risk customers across Google and Meta.
The results didn’t come from a single campaign. They came from a rhythm. Every other week, the Big Dee’s merchandising and creative team sits down with the LXRInsights growth team to read the data, decide what to test next and agree on who brings what.
Customer trajectory, churn risk and experiment results from LXRInsights.
Pick the next hypothesis, campaign structure and budget split.
Big Dee’s curates SKUs and creative. LXRI ranks products by past performance.
Each test runs against a control campaign so wins are real, not guessed.
Winners scale. Everything else feeds the next hypothesis, logged in a shared living document.
Early sprints mapped the business: competitor customer acquisition, high-churn retention, peak-hour targeting, high-margin products and a series of high-potential market tests.
Competitor acquisition scaled first at 4.37x ROAS. The market tests taught the team which regions to lean into and which to leave alone, and that learning set up the geo wins that came later.
With audiences better understood, the team tested which product groups to push. Horse care (7.61x), stable products (7.45x), low-visibility products (7.45x), customer quality (7.37x) and high-intent keywords (4.32x) all beat control and scaled.
LXRInsights predictive audience plays moved to the center: high-value seed audiences for PMax, lookalikes of the best customers and win-back lists for customers likely to churn.
An experiment succeeds when it’s repeatable enough to run permanently, and a losing test is cheapest when it’s caught early.
A test that lifts orders or clicks can still lower order value or ROAS. We judge every test on business outcomes.
Daily numbers swing. We wait for statistical confidence before scaling or stopping a test.
Cheaper clicks don’t always mean better customers. High-value seed audiences teach the algorithms who matters most.
Predictive audiences amplify real demand. Products that have been dormant for years usually stay that way.
Bringing lapsed and at-risk customers back was Big Dee’s single highest-ROAS play.
Big Dee’s and LXRInsights were named finalists in MediaPost’s 2026 OMMA Awards for AI-powered experimentation that turned customer intelligence into growth.
