Case Study: Propensity-to-Purchase Modeling Unlocked Hidden DSL Demand

Rural and smaller markets across the West and Upper Midwest lagged DSL adoption compared to peer regions. Traditional channels like radio, billboards, and broad direct mail were costly and underperforming, and privacy limits restricted access to household demographics or ARPU.

Current customers with specific bundles of telco services were far more likely to add DSL. Using the mix of existing subscribed services as the signal, you could predict who would convert without needing sensitive personal data.

• Aggregated a 300k-plus account dataset across WA, CO, MI, WY, NE, NC, SC, LA, and TX to find service combinations correlated with DSL ownership.
• Scored the base and produced a 100k-target list of active customers who looked DSL-likely but hadn’t subscribed yet.
• Activated only low-cost channels to validate the hypothesis: email, direct mail, and bill stuffers, holding back radio and billboards to protect CAC.

• Targeted campaigns delivered 12 to 19 percent uptake in month one, versus 1 to 3 percent for typical mailers.
• Broadband revenue grew an additional 6 percent in year one, about $1.1 million in new revenue.
• Other product lines adopted the same modeling approach after seeing the lift.

• Don’t default to price cuts. Mine your existing customer base first and use owned signals to predict demand.
• Prove lift with low-cost channels before scaling media spend.

Impact statement: A simple, privacy-aware propensity model turned stagnant rural markets into high-yield targets and created a repeatable playbook for other product teams. Structured to match your Better Together case study.