Replenishment Is Not Logistics. It’s Revenue Protection.

QUICK ANSWER. A stockout is not a logistics failure. It is a revenue event — the moment a customer walks past your empty machine and buys coffee somewhere else. Manual ordering and fixed delivery schedules produce those moments by design. Consumption data should trigger the delivery instead.

Most operators file replenishment under logistics. Routes, vans, pallets, a delivery frequency agreed once at contract signing and rarely revisited.

It belongs in the revenue column.

Every bag that arrives late is a cup bought somewhere else — or not bought at all.

What actually happens when a machine runs dry?

Nothing dramatic. That’s the problem.

Nobody escalates a stockout the way they escalate a broken machine. The office finds a workaround within the hour: the café downstairs, a jar in the kitchen, the competitor’s machine on floor three. Consumption doesn’t pause when your coffee runs out. It relocates.

And workarounds are sticky. Once a team has found a route to coffee that works, it tends to survive the refill by several weeks.

“A stockout doesn’t cost you a delivery. It costs you the habit.”

This is the same leak we described in Revenue Leakage: The Coffee Sales You Lose Without Noticing — with one difference: authorship. In classic leakage the customer chose another supplier. In a stockout, you sent them.

Why does manual ordering keep failing?

Because it asks people to predict something none of them can see.

• The customer orders when someone notices the hopper is low — which is already too late for a delivery slot next week.
• The account manager orders from history — what this site took last quarter, adjusted by feel. Offices don’t consume by quarter. They consume by headcount, season, and whatever landed in the calendar that week.
• The fixed schedule delivers the same quantity every four weeks, whether the machine served four hundred cups or four thousand.

All three fail in the same two directions at once: emergency deliveries to sites that ran out, and full cupboards at sites that didn’t. One costs margin in expedited logistics. The other ties up product and hides the real consumption picture behind a stock buffer nobody measures.

Scale sharpens it. Across a European vending and office coffee service market that EVA, the European Vending & Coffee Service Association, sizes at roughly 4.5 million installed machines across 24 markets, replenishment is one of the highest-frequency decisions in the entire industry — and one of the least instrumented.

What should trigger a delivery instead?

Three signals the fleet already produces every day:

1. Consumption rate. Cups served per product, per site, since the last delivery. Not an estimate — a count.
2. Days of stock remaining. Consumption rate measured against what was actually delivered gives a burn-down. A burn-down gives a date.
3. Route and account context. Which sites sit on the same run, which contracts carry service-level exposure, which accounts you cannot afford to leave empty on a Monday.

Together they turn replenishment from a calendar into a queue: which site runs out first, and what has to be on the van to stop it?

The forecasting logic behind that queue — and the KPI set that keeps it honest once it’s running — is Step 5 of the 9-step operating model in our office coffee service report. The method is in the report, deliberately not in this article.

Isn’t automated replenishment just handing control to a machine?

No, and the distinction matters more than it sounds.

CoffeeBrain is designed to be read-only and pull-only: it reads consumption from the installed fleet and raises a trigger. The trigger lands in the operator’s own ERP and logistics plan, where a human confirms it. Nothing is ever written back to a machine, and no machine orders its own beans.

Automated deliveries means the decision arrives early, with the evidence attached. The operator still makes it.

What changes for the sales side?

Sales stops learning about consumption at the annual review.

When cups served and product delivered sit in the same view, an account that is quietly consuming more than it buys becomes visible in weeks rather than at renewal — and an account whose volume is sliding becomes a conversation while it can still be saved. That’s the same visibility that decides which machines actually make you money, applied to the product rather than the machine.

It also compounds with service. A van already routed by need — see Service the Machine When It Needs It — is a van that can carry the right product to the right site on the same trip.

Key takeaways

• A stockout is a revenue event, not a logistics inconvenience — consumption relocates, and workarounds are sticky.
• Manual ordering and fixed schedules fail in both directions: emergency deliveries and overstocked sites.
• Replenishment should be triggered by consumption rate, days of stock remaining and route/account context — not by a date.
• Automation here means a trigger routed to humans and the ERP, never a command sent to a machine. Read-only, pull-only.
• The forecasting logic and KPI set are Step 5 of From Coffee Machines to Profit Machines.

How many of last month’s deliveries arrived because a date came up — and how many arrived because a site was about to run out?
Request the white paper — From Coffee Machines to Profit Machines: a step-by-step model for a more profitable OCS operation.