‹ All work
Web app,
Haulpoint,
2026
Route planning in 9 minutes, not 40
A planning board that starts from last week, so dispatchers only plan what changed.
Company
Haulpoint
Role
Lead product designer
Team
A PM, six engineers and a data analyst
Timeline
Six months, 2025-26
Platform
Web app
The short version
1 min
Problem
Dispatchers spent about 40 minutes every morning planning routes, across three screens and a spreadsheet.
What I did
Shadowed six dispatchers at three depots, then designed one planning board that starts from last week’s plan and shows only what changed.
Result
Median planning time fell from 40 to 9 minutes across 31 pilot depots, and 24 of them stopped using their spreadsheet.
My role
Lead product designer. Research, flows, UI, copy and the spec.
Context
Haulpoint sells delivery software to regional couriers, companies running 10 to 80 vans out of a few depots. Every morning before the vans leave, a dispatcher assigns the day’s orders to drivers and routes.
The old planner split that job across three tabs: orders, a map and a driver list. Nothing told you a driver was over their hours or a van was full until you saved the plan and the error came back. Most dispatchers had given up. They planned in a spreadsheet, then typed the result into Haulpoint.
My role
Lead product designer, from the first depot visit to the rollout. I planned the research with our data analyst, owned every flow and screen, and wrote the words on them. During the pilot I sat with the engineers twice a week, so the spec never fell behind the build.
Research
Before touching the old planner, we went to where planning happens. Our analyst and I spent two mornings at each of three depots, next to six dispatchers, from 5:30 a.m. until the last van left.
The spreadsheet wasn’t a workaround for a missing feature. It held something the product didn’t: last week. Dispatchers copied last Tuesday’s tab and changed the few routes that were different. Our analyst then checked six months of plans: on a typical day, 4 in 5 routes were almost the same as the same weekday a week earlier.
6
Dispatchers shadowed
3
Depots visited
6
Mornings at 5:30 a.m.
“I don’t plan the day. I plan what’s different from last Tuesday.”
Dispatcher, 34 vans
Plans repeat.
4 in 5 routes match the same weekday last week.
Errors arrive late.
Hour and capacity limits only showed up after saving.
The map is for checking.
Dispatchers looked at it last, to confirm a route, not to build one.
Approach
The same six dispatchers tried three directions, on paper first and then as clickable prototypes:
Map first
Like most competitors. It looked great but was slow: dispatchers dragged pins for 20 minutes.
A table
Like their spreadsheet. Fast, but they lost track of which van was full.
A board
Drivers as columns, stops as cards and last week’s plan already filled in. It was the fastest in every session.
That set three rules: one screen, start from last week, check limits as you go.
Solution
The board opens on today, already filled in with the plan from the same weekday last week. Orders that are new or cancelled since then sit in a tray on the left, marked as changes.
Each driver is a column. Dragging a stop shows its effect right away. A column turns amber when a driver gets close to their hours limit and orange when a van is over capacity, with the reason in plain words.
The map moved into a side panel that follows whatever you select. The labels now use the words dispatchers say on the phone: “Unallocated consignments” is “New orders”, and “Shift violation” is “Over hours by 25 min”.
The board: last week’s plan, with today’s changes in the tray.
The old planner: orders, map and drivers in three tabs.
The board: one screen, starting from last week’s plan.
A van over capacity, with the reason shown in its column.
Result
Median planning time
24 of 31
Depots retired their spreadsheet
−52%
Planning tickets in three months
The board went live at 31 pilot depots in February 2026, and for every customer in May.
The one problem still left is changes during the day, when a van breaks down after it has left. That’s the next project.
What I learned
Watch the work before you redesign the screen. The feature dispatchers thank us for is a copy of last week, and no amount of polish on the map would have found it.
Bring the data analyst in early. The 4-in-5 number turned a hunch from six mornings into something the whole company was willing to build on.
Next project
Card disputes customers can finish alone →
Kitebank,
iOS,
2025
