Redesigning pricing buckets at a top LATAM marketplace
+11.5% inbound, -2.8pp contact rate, -2.4pp bad stock
The problem
The marketplace's fee structure had evolved organically over several years, resulting in pricing buckets that no longer reflected the actual cost structure of different product categories. Sellers in some categories were over-indexed on fees relative to their margin potential, suppressing supply growth. Others were under-indexed, creating adverse selection where low-quality listings dominated certain segments.
The result: declining inbound in key categories, rising contact rates (a proxy for buyer dissatisfaction), and growing bad stock — items listed but never sold, cluttering search results and degrading the buyer experience.
What I did
Mapped the fee-to-margin landscape across 200+ categories
Built a comprehensive model connecting fee tiers to seller economics across every major category. This wasn't a spreadsheet exercise — it required synthesizing data from pricing, category management, and seller operations teams to understand where fees were misaligned with the value delivered to sellers.
Designed a rebucketization framework with guardrails
Created a new bucket structure that realigned fees with category margin potential while maintaining revenue neutrality at the portfolio level. The key insight was that simply lowering fees in suppressed categories wouldn't work — we needed to simultaneously adjust adjacent buckets to prevent arbitrage and maintain seller trust.
Coordinated rollout across 5 teams with zero downtime
The change affected millions of active listings. Rolled out in phases with monitoring dashboards, automatic rollback triggers, and seller communication sequences. Every phase was validated against pre-defined success criteria before proceeding.
What I learned
Pricing changes at marketplace scale are fundamentally different from feature launches. The blast radius is immediate and universal — every active seller is affected simultaneously. The most important skill isn't analysis; it's building confidence among stakeholders that the change won't cause a crisis. That confidence comes from rigorous scenario modeling and clearly communicated rollback plans, not from perfect predictions.
Numbers
+11.5% inbound in re-bucketed categories (approximate, anonymized)
-2.8pp contact rate reduction
-2.4pp bad stock reduction
Revenue-neutral at portfolio level within 60 days
Numbers are approximate and anonymized where applicable.