Middleman
A restocking app for a beer route
I ran a daily beer route at New Hampshire Distributors, restocking grocery and convenience stops at five in the morning and ordering with a pen on cardboard. Middleman is a prototype of that job done on a phone, with the shelf count, the pull list and the order in one pass.
Try the prototype



Stop
The first question in any store is which SKUs moved since the last trip. This screen answers it before you reach the aisle.
Three stock counts
Critical, warning and healthy, each with its own number, and nothing else sharing the row with them.
Four action tiles
Notes, pull, delivery and breakage, the delivery tile carrying its ETA. Those are the four things you actually do in a store.
Activity list
Auto-orders, promos, delays, pulls and deliveries, with a name against anything a person did rather than the system.

Critical, warning, healthy

Four jobs, one row

What moved since last visit
Shelf
Counting a store properly took two and a half hours. Most people hit it fast and missed cases instead. The count is on the screen before you start.
Stock row
The SKU, two chips flagging backorder and critical, the shelf count against maximum, days of supply left, backstock and how full the facing is.
Three floor buttons
Pull list, breakage and shrinkage, because those are the three reasons you open this screen.
One SKU, expanded
Stock against maximum, the reorder point and the case size, so the pull is a number of cases rather than a guess.

One row, the whole shelf

Pull, breakage, shrinkage

Stock, reorder point, case size
Order
An order went in on paper and came back as a truck. Everything between those two moments was invisible.
Delivery tile
The ETA with how far along the run it is, the case count, a note that reaches the driver and the full order behind it.
Order in progress
Still building at a running case count, with the last change and the name on it, and the whole order one tap from editing.
Line controls
Every line on an open order takes a plus, a minus or a delete, for as long as the order stays open.

On the truck now

Still open, still editable

Edited by the case
Route
Seven steps, every store, every day. Five of them ran on memory, and none of it was written down anywhere.
Pen and cardboard
Counting one store properly took two and a half hours.
Middleman
The count is on the screen before you walk in.
Walk in with no numbers
Open on the store you are in
Count the shelf by eye
The shelf count is already there
Tally it on cardboard
The pull comes out in cases
Pull from the back from memory
Backstock sits on the same row
Hand the order over on paper
The order is already building
Breakage goes unlogged
Log breakage where it happened
Variance turns up months later
Shrink is tracked per SKU

Five of the seven happen on this one screen.
Impact
Nothing here shipped. It is the daily beer route I worked, drawn the way it should have been built.
2.5 hrs
To count one store properly
9
Screens, each with its own route
0
Tools the job actually issued
Designed from inside the route rather than from research. I worked it.
Learnings
Design what you have done
Working the route is why these screens answer the real question instead of the one a brief would have asked.
Cut the camera idea
Photographing the shelf still left you counting it. The goal was knowing the numbers before you walk in.
Density reads better on a cart
A terminal layout looks hostile on a desk and reads perfectly on a cart in a cold aisle.
MORE WORK