Case studies

Same budget. More revenue. Proven against a control group.

Two client engagements where AI-powered targeting replaced blanket campaigns, and the lift was measured, not claimed. Client names withheld on request.

Case study 01 · Large telco operator

AI-driven next-best-offer, measured on every campaign.

A telco operator running dozens of concurrent recharge, data-pack and upsell campaigns across SMS, app-push and voice. Campaigns were high-volume and low-precision, the same offer to millions of subscribers, no way to tell what actually worked.

2.4x

revenue per contact

+18%

incremental ARPU on targeted base

40%

reduction in wasted sends

What we built

A per-subscriber propensity and value engine that scored the entire base daily against a library of offers, and picked the right offer, right channel and right moment for each subscriber.

How we measured it

Every campaign shipped with a randomized holdout. Reported lift is treatment minus control on the same period, not year-over-year, not vs a soft baseline.

What changed operationally

Campaign teams stopped guessing target segments. The system ranked the eligible base, capped fatigue across channels, and sent a weekly report showing which offers moved the needle and which were quietly losing money.

Case study 02 · Consumer e-commerce brand

Cutting the send list. Growing the revenue.

A consumer e-commerce brand blasting weekly promotions to its full customer base. Open rates were falling, unsubscribes climbing, and nobody could say whether the campaigns were actually driving revenue or just harvesting demand that would have converted anyway.

1.7x

revenue per targeted customer

−60%

send volume

+22%

campaign-attributable revenue

What we built

A ranked customer list refreshed weekly: who is likely to buy in the next 30 days, what they're likely to buy, and how much they're worth over the next six months. Champions were ring-fenced, lapsers got a targeted win-back, and low-value one-timers were suppressed.

How we measured it

Every send was split into treatment and a randomized holdout of the same shape. We reported incremental revenue, the revenue that would not have happened without the campaign, not gross attributed revenue.

What changed operationally

The team stopped sending to their whole list. Fewer sends, higher revenue per send, and a weekly dashboard that made it obvious which campaigns were incremental and which were pulling forward sales.

Work with us

Ready to see the same lift on your data?

Your first-party data is the only real moat. We work on it directly, with you, not generic AI, not chatbots, not off-the-shelf models. One measured campaign against a randomized holdout, and you keep the lift.