Apurv Patil

• 8-figure USD — Annual Revenue Gain from Supplier Routing Optimization
• $1+ Million — Monthly Booking Revenue Protected Through Automated Supplier Failover

Chase Travel sources air inventory across several suppliers, and that layer had no decision logic. Traffic went where it always had, so cardholders sometimes saw worse fares than the same itinerary sold elsewhere, and Chase left commission behind. Worse, when a supplier failed mid-transaction, the booking died at checkout, even when the identical itinerary sat available with another supplier.

I treated routing and failure as one problem.

Routing: I mapped supplier economics against fare competitiveness by route and cabin, then defined logic to shift volume only to suppliers winning on both. Guardrails prevented commission optimization from ever degrading a cardholder’s fare.

Failover: I specified a mechanism that detects a mid-transaction failure and reroutes to an alternate supplier holding the same itinerary, preserving fare integrity. The design work sat in the edge cases—retry criteria, acceptable price delta, and double-booking prevention on late confirmations.

Supplier selection became a live commercial lever instead of a static configuration. Routing delivered 8 figures in incremental annual revenue, and failover recovered a 7-figure monthly revenue line that was previously abandoned at checkout.

• 80% — Reduction in Time to Diagnose Metric Drops
• 40% — Increase in Weekly Active Usage Among Merchant Founders & CXOs

Every analytics dashboard in the category answers what happened; none answer why. A founder watching revenue fall on a festival day could see the number, but tracing it back—revenue fell because AOV fell, AOV fell because a top campaign stopped converting—took an analyst and several hours. On the days that mattered most, nobody had hours.

I built a deep-dive analytics product that automatically performs causal decomposition and returns the results in plain language.

Instead of surfacing another chart, the system walks the metric tree from the headline number down through its drivers, isolates the ones that actually moved, and uses an LLM to explain the chain in the terms a founder thinks in. The product calls that mattered: scoping which metric relationships to model, setting the bar for when the system stays silent rather than asserting a weak cause, and designing output a CXO can act on in under a minute.

Founders diagnose performance changes themselves, in minutes, without an analyst—most valuably on peak event days when the window to react is hours. It moved the platform from a reporting tool to a decision layer.

• 30% — Increase in Premium & Business Cabin Availability
• 18% — Lift in Business Class Booking Conversion

Cardholders searching for business and first class routinely found nothing available. The cause was structural: each supplier was queried for a complete round trip, so if no single supplier held inventory in both directions, the search returned empty—even when the outbound sat with one supplier and the return with another. Premium cabins are where the highest-value cardholders spend.

I specified a fare construction approach that assembles itineraries across suppliers rather than within them, pairing legs from different sources into one bookable trip.

The search was the easy part. The real work was downstream: ticketing across 2 PNRs, servicing a trip whose legs sit with different suppliers, and owning disruptions when one leg fails. I worked those paths through with operations before writing requirements, because an itinerary that cannot be serviced is worse than one never offered. I also set ranking logic for when a mixed fare outranks a single-supplier one.

Searches that previously returned nothing now convert. Premium cabin availability expanded materially for the cardholders whose spend justifies a premium card, without adding a single new supplier contract.

• 34% — Increase in Post-trial Retention
• 22% — Increase in Trial Activation Rate

Merchants across India and Southeast Asia signed up for trials, finished onboarding, then stopped logging in. Conversion to paid hinged on whether a brand reached a first genuinely useful result within the product, and most never got there. The drop-off was not in acquisition—it was the 2 weeks after setup, and CS was diagnosing it on anecdote.

I instrumented the trial funnel to find where merchants actually stalled rather than relying on customer success anecdotes, then rebuilt onboarding around reaching a first meaningful output faster—cutting setup steps, automating what did not need a human, and redefining what CS owned in week 1.

I ran it as product owner across 2-week sprints, with activation and retention set as sprint-level success criteria rather than metrics reviewed quarterly.

Post-trial retention rose by around 34% and activation by around 22% across Southeast Asian markets, moving the growth constraint off acquisition spend and onto product experience, where it was far cheaper to fix.

• 1st — US Dealer Pipeline Built From Zero at IBS 2026
• 15% — Unlevered IRR Base Case Clearing on a 10-year Entry Model

A precision manufacturer with no US presence needed to enter a mature market in which incumbents control the dealer channel. No brand recognition, no distribution, no pricing reference—and a product where the buying decision sits with dealers rather than end customers. The entry case had to hold up on numbers before any capital moved.

I built the commercial case and entry mechanics end to end: a 10-year financial model covering unit economics, capex phasing and levered returns, and a 7-stage dealer acquisition funnel with a 5-touch outreach sequence and founding-dealer pricing structured to buy early distribution cheaply.

I then took it to market at the industry’s largest US trade show and tested the positioning against real dealers rather than validating it in a deck.

A US entry model validated in the market rather than on paper, with a dealer pipeline built from zero and a return case that cleared the threshold before capital was committed.