Data-Driven Push-Sale at FPT

Four months at FPT Corporation, between Garena and Ninety Eight. February to May 2023. I knew going in this was a bridge. Took it for the money jump, left as planned.

The Brief

The team was running customer acquisition with manual instinct: spend where last quarter spent, double down on what felt right. No causal read on which channel, which creative, which segment was actually moving CAC. The dashboards were rear-view; the decisions were vibes.

I rebuilt the read.

What I Shipped

A data-driven push-sale system. Three pieces:

• Attribution wired from ad spend to first purchase, segmented by channel + creative variant + audience.
• Weekly review cadence against a rolling cost-per-acquisition target, with the ledger visible to BD, marketing, and finance in the same view.
• Kill / scale rules so weekly decisions stopped depending on whose turn it was to argue.

The numbers stopped being decoration and started being signal. Channels that looked busy but weren’t converting got cut. Creative variants that compounded got more budget. CAC came down 18 percent over the four months I was there.

The Honest Frame

Not a scale event. A clean, scoped, one-system contribution at a corporate-driven org that wasn’t my long-term fit. I left for Ninety Eight when the next move was clear. The system stayed.

This is the bridge role on the arc, not the proof. The proof lives at Garena (5 SEA countries, 2M+ followers), Ninety Eight (0 to 1.5M MAU in 90 days), and Kyber (the $200B+ funnel pipeline).