Operations Management When You Run Your Own Deliveries

Operations management is the work of designing and running the process that turns inputs into finished orders: scheduling it, stocking it, checking it, and sizing it to what you can produce. In a textbook it is a factory discipline. In a bakery with a van, a florist with two drivers or a caterer running four events a weekend, it is the same discipline compressed into a day, with a hard deadline attached to every unit: somebody’s delivery window.

That deadline is what makes operations management different for a business that delivers its own orders. A retail shop can finish a batch late and sell it an hour later. You cannot. The truck leaves, or it doesn’t. Everything below is about building a process that survives that constraint. For the wider view of how this function sits alongside sourcing, money and people, start with the guide to business operations for a shop that delivers.

The Bottom Line

• Schedule production backward from delivery windows, not forward from opening time. The route is the constraint; the kitchen or workbench is the variable.
• Inventory errors are the most expensive quiet failure in the business. IHL Group put the global cost of inventory distortion (out-of-stocks plus overstocks) at $1.77 trillion in 2025, with out-of-stocks accounting for $1.2 trillion of it.
• Quality control has to happen before loading. A defect caught at the bench costs minutes; the same defect caught at the customer’s door costs a redelivery and the margin on the order.
• Capacity is measured in the narrowest window of your week, not in your weekly total. Most shops are at 60% on Tuesday and 130% on Friday and call it a staffing problem.

What an operations manager actually does in a small delivery-heavy business

In a company with a dedicated operations manager, the job is usually described as process design, resource allocation, scheduling, inventory, and quality. In a shop where the owner is the operations manager, the same responsibilities exist but get done in fragments between other work, which is why they are the first things to slip.

The four that matter most when orders leave on a vehicle:

• Sequencing the work so items are finished in the order they are loaded, not the order they were sold.
• Holding the inventory buffer on the few inputs that stop everything when they run out, and refusing to hold it on everything else.
• Deciding what the quality gate is and where in the process it sits, which has to be before the item is on a truck.
• Setting the cutoff for how late an order can be accepted for a given day, and enforcing it.

Notice that three of the four are decisions made once and then defended, not tasks performed daily. That is the useful reframe: most operations management in a small business is deciding a rule and then not making exceptions to it at 6 a.m. on a Friday.

Production scheduling against fixed delivery windows

The default schedule in most small shops is product-shaped: make all the sourdough, then all the rye, then all the baguettes. It is efficient at the bench and inefficient at the door, because it finishes the entire day’s output at roughly the same time and leaves the first van waiting.

Schedule route-shaped instead. Build the routes first, or at least the drop groups, then work backward:

1. Group the day’s orders by route and by window. East side 9–11, west side 11–2, and so on.
2. Set a load time for each route, working back from the first drop plus travel.
3. Set a finish deadline per group, typically 30 to 45 minutes before load time to allow packing and checks.
4. Schedule production in that order, accepting some batching inefficiency in exchange for vehicles leaving on time.
5. Leave one buffer slot per session for the order that gets added late or the batch that fails.

Name the trade before you make it: route-shaped scheduling costs you some changeover time at the bench. It buys you on-time departures, which is what customers notice and what costs money when it fails. MIT Sloan Management Review’s work on last-mile costs makes the same point at enterprise scale, finding the expensive failures concentrated at the delivery end rather than in production efficiency (MIT Sloan Management Review).

The other half of scheduling is the cutoff time, and it is mostly a discipline problem rather than a planning problem. An order accepted after the cutoff does not just add work; it re-sequences work that has already been planned, and the cost lands on every other order in that session.

Inventory control for a business that delivers

Inventory management in a delivery operation has one extra requirement that a counter business doesn’t have: you have to know you can fulfill an order before you promise a window for it, because a stockout discovered at 5 a.m. is not a stockout, it is a cancelled delivery.

The scale of the problem is not small even at large operators. IHL Group, which has tracked the measure for nearly two decades, put global inventory distortion at $1.77 trillion in 2025, $1.2 trillion of it from out-of-stocks and $572 billion from overstocks, equal to about 6.5% of global retail sales (IHL Group). Both halves of that number hurt a small shop more, because you finance the overstock out of working capital and you eat the out-of-stock as a failed delivery.

A workable inventory control process at this size has three parts:

• Classify inputs by what they stop. A handful of items halt production entirely when they run out. Those get a real buffer and a reorder point. Everything else gets ordered on a normal cycle, and running thin on them is fine.
• Count what moves, on a cycle. Counting a small subset weekly catches drift far earlier than counting everything twice a year, and it takes fifteen minutes rather than a Sunday.
• Reconcile against orders, not against the shelf. The question that matters is not “how much do we have” but “how much do we have against what is already promised for Thursday.”

Perishable inputs make this sharper rather than different. The buffer that protects a delivery window also spoils, so the reorder point has to be set against shelf life, and the honest answer for some items is a smaller buffer plus a supplier who can deliver twice a week.

Building a quality control process that runs before the van leaves

Quality control in a shop that delivers has to move upstream, because the customer inspects the product without you there. There is no counter conversation, no swap, no recovery. There is a photo and a phone call.

Three checkpoints cover most of it:

At production. The person who made it checks it against a stated standard, not a vague one. “Iced evenly, no gaps at the base” is a standard. “Looks good” is not.

At packing. Check the order against the ticket: right items, right counts, right customer. Order-accuracy errors are far more common than product-quality errors and far more annoying to the recipient, and they are caught entirely at this step.

At loading. Check that what is on the vehicle matches the route manifest, in reverse drop order. This is the checkpoint most shops skip, and it is the one that prevents the worst failure mode: a correct order delivered to the wrong stop, which costs you two customers instead of one.

Keep a one-line record of every failure caught and every failure that escaped. After a month the list stops being a record and starts being a to-do list, because the same two or three causes will account for most of it.

Capacity planning: how much can you actually deliver in a day?

Capacity planning at this size fails for one specific reason: owners calculate it weekly when the business runs in sessions. You do not have a 50-order week. You have a 6-order Tuesday and a 22-order Friday, and the Friday is what defines your real capacity.

Work it out per session, and take the smaller of two numbers:

• Production capacity: how many units the bench can finish between the earliest practical start and the finish deadline for the last route.
• Delivery capacity: how many drops a vehicle can complete inside the promised windows, including load time, travel between stops, parking and time at the door.

Whichever is smaller is your capacity. Adding staff when delivery is the binding constraint just produces finished goods that sit; adding a route when production is the constraint produces an empty van. This is the same constraint logic that governs the whole operation, and it is worth rechecking every time volume steps up.

Two numbers make the plan honest, and both are commonly wrong by a wide margin in owners’ heads: actual minutes per drop, door to door, and actual changeover time between batches. Time them for a week. Whatever you currently assume, the real figures are usually longer, and every schedule built on the optimistic version fails on the busiest day, the one day it mattered.

What operations management can’t fix

Process work has a boundary. If each order loses money once delivery cost is counted, a better schedule produces the same loss more efficiently. That is a pricing and cost question rather than a scheduling one, and it gets settled by the numbers on the financial side of a delivery operation: cost per order, margin per route, and how long the money takes to arrive.

The practical division: if your week is chaotic but the orders make money, you have an operations management problem and this is the work. If your week runs smoothly and you are still short at the end of the month, no amount of sequencing will fix it.

Start with one session

Pick your busiest session. Build the routes first, set the load times, and schedule production backward from them for two weeks. Track two things only: the time each vehicle actually left, and the number of orders that needed a phone call.

Those two numbers will tell you within a fortnight whether the process change held. Most shops find the departure times tighten before the phone calls drop, because sequencing fixes timing first and accuracy second. Both improve, in that order.