The Last Mile Is the Whole Game

The hardest problems are the ones you can't see from a desk

Last-mile logistics looks tidy on a dashboard and chaotic on the ground. A planner draws a clean route; the driver hits a closed road, a gate with no buzzer, a customer who isn't home. The gap between the plan and the street is where most delivery products quietly fail.

So I don't trust the happy path. The best product decisions I've made came from sitting with the people who actually live in the system, dispatchers reassigning a stop at 9am, drivers squinting at an address that doesn't exist, planners trying to absorb a no-show. If a feature doesn't survive contact with a bad traffic day, it isn't done.

Optimize for the people, not just the route

I spent years on route optimization and sectorization before I moved into product, and that background taught me something the math alone can't: the most efficient route is useless if the driver doesn't trust it. An algorithm that saves three minutes but ignores a known shortcut, or reorders stops in a way that feels wrong on the road, gets abandoned on day two.

A delivery system has several users with different anxieties, dispatchers want control, planners want predictability, drivers want to get home, customers want to know when. You can't serve all four with one screen. The job is to reduce friction for each of them without optimizing one into the corner of another.

Data points the way, the field confirms it

I lean hard on product usage data, A/B testing, and NPS to decide where to invest, it keeps me honest about what's actually working versus what we hoped would. But data tells you *what* is happening, rarely *why*. A drop in completed deliveries is a number; the reason is a confusing reassignment flow, and you only learn that by watching someone use it.

So I pair the two. The data finds the bleeding edge; the customer interview explains it. Neither alone is enough to commit a roadmap to.

What I keep coming back to

Delivery is unglamorous and unforgiving, and that's exactly why I like it. There's no hiding behind a nice demo when a package is late. Build the tools that make a dispatcher's morning calmer and a driver's route make sense, and the efficiency, scalability, and sustainability everyone wants downstream tend to follow. Get the last mile right and you've earned the whole journey.