How I think about building products
The best products don't start with interfaces — they start with operations. Understand how work actually flows before you design how it should flow. Technology should encode hard-won operational knowledge, not replace it with assumptions.
Principles
Spend time in the warehouse, ride along with the driver, watch the operator's workaround. The real product requirements live in the gap between the current process and the ideal one.
At S.F. Express, field observations of driver routing revealed that 30-40% of time was spent on failed delivery reattempts — directly shaping the smart locker network strategy.
You can't improve what you can't measure. Before building features, build the data layer. Dashboards and analytics aren't afterthoughts — they're prerequisites.
At FoodsUp, Tableau and Metabase dashboards analyzing supplier performance and pricing trends drove a 20% improvement in procurement decision-making.
A feature solves one problem. A platform creates the conditions for many problems to be solved. Design for the ecosystem — connect operators, partners, and users through shared infrastructure.
The smart locker network grew from a single delivery solution to a multi-sided platform connecting logistics providers, merchants, and consumers across 2,000+ locations.
Digital transformation fails when it ignores the humans in the loop. Stakeholder buy-in, change management, and training are product work — not afterthoughts.
At Kowloon Motor Bus, launching the bus captain app required stakeholder meetings across 5,000+ captains to ensure seamless adoption, reducing operational expenses by 30%.
Process
Embed in operations. Interview users, observe workflows, analyze failure data. Understand the problem before proposing solutions.
Test assumptions with data. Run pilots, build prototypes, measure against operational baselines. Kill ideas that don't survive contact with reality.
Ship in phases. Start with the hardest constraint, instrument everything, iterate with real users. Cross-functional teams, Agile cadence.
Track outcomes, not output. Did we move the operational metric? Did we reduce cost, time, or errors? Dashboards tell the truth.
Expand to new markets, new use cases, new partners. Localize features, build platform APIs, let the ecosystem grow.