Found hidden cost leakage in provider payouts and non-completed sessions, recovering ~10% of operational costs.
As Intellect scaled across 36 countries with 500+ providers, operational costs were growing faster than revenue. No one had done a deep audit of where money was actually going. Three separate leakage areas were eventually uncovered:
High-cost providers being recommended when affordable alternatives existed.
Users missing recurring sessions but providers still getting paid.
Non-completed sessions still counted in payout calculations.
Operational costs weren't just growing with scale — there was structural leakage in how sessions were being allocated, completed, and paid for. A data-driven audit of the payout pipeline would reveal recoverable costs without reducing service quality.
Built a comprehensive cost analysis tracking every session from booking to payment. This audit uncovered three distinct leakage areas, each requiring a different fix.
Found a S$457K savings opportunity, representing 16.5% of base costs. High-cost providers were being surfaced when 48% of supply was affordable. Recommended a cost-penalty layer on the recommendation algorithm.
Found that users missing one session had a 75% probability of missing ALL subsequent ones, yet providers were paid for all. Ran a backend experiment requiring user confirmation for upcoming sessions.
14.4% of total spend was going to sessions that never completed. Proposed shifting from credit-based to session-based payouts, recalculating bonuses only for completed sessions.
Recovered significant operational spend that was being wasted on non-value activities.
Changed how the company thinks about payout models — from "pay for allocation" to "pay for completion."
The cost-penalty layer improved recommendation quality while also reducing costs — a rare win-win.
Finance team now has real-time visibility into cost leakage through dashboards built from this analysis.
Looking for a data person who can go from SQL to boardroom? I'd love to chat.