Capacity Planning for Delivery Operations: Matching Resources to Demand

Most operators find out they’re at capacity the hard way. A big order lands, the van is already full, and someone spends an hour calling customers to move delivery windows. Nobody planned for that day. It just arrived.

Capacity planning for delivery operations is the work that stops that phone call from happening. It answers one question in advance: how many orders can we actually get out the door in a day, and what happens when demand goes past that number? Everything else is downstream of the answer: hiring, buying a second vehicle, taking on a wholesale account, saying yes to a holiday rush.

This guide covers the whole cycle. How to measure your real capacity, how to forecast what’s coming, how to find the one constraint that’s actually holding you back, and how to choose between adding capacity early, adding it late, or adding it in steps. Two pieces of the picture have their own guides: the physical warehouse layout that determines how fast orders get picked, and what to do when your suppliers are the thing that breaks.

The Bottom Line

• Capacity is a number, not a feeling. Measure orders shipped per day at each stage (picking, packing, loading, driving). The smallest number is your real capacity.
• One stage is always the constraint. Adding vans when the bottleneck is packing buys you nothing but a parked van.
• The last mile is where the money goes. Capgemini Research Institute’s work on last-mile economics put final-mile delivery at roughly 41% of total supply chain costs, and more recent industry benchmarks put it above 50% in dense urban operations. Capacity decisions here move your cost line more than anywhere else.
• Pick a strategy on purpose. Lead, lag, and match are three different bets on demand. Choosing by default usually means lagging, which is the one that costs you customers.
• Plan on two clocks. Weekly adjustments handle normal swings; six-month-plus decisions handle growth and seasonal peaks.

What Capacity Planning Means in a Delivery Operation

Capacity planning is the process of comparing what your operation can produce and deliver against what customers are going to ask for, then closing the gap before it opens. In a manufacturing plant that means machines and shifts. In a business that delivers its own orders, it means something more specific: the number of orders you can pick, pack, load, and physically drive to a customer inside a working day.

That last part is what makes delivery capacity different from plain production capacity. A bakery can bake 400 boxes of pastry. Whether it can deliver 400 boxes depends on vehicles, drivers, route density, traffic, and how long a driver stands at each door. Baking capacity and delivery capacity are two different numbers, and the smaller one is the one that governs what you can sell.

The stakes sit in the cost line. Last-mile delivery has become the single most expensive segment of the supply chain. Capgemini Research Institute’s research on final-mile economics is the source most of the industry still cites for the 41% figure, and later industry benchmarks put the share higher in dense urban routes. When delivery is half your logistics cost, running it 20% over capacity or 20% under capacity is an expensive mistake in both directions.

How to Calculate Your Delivery Capacity

Start with a stopwatch, not a spreadsheet. Delivery capacity is the throughput of the slowest stage in your order flow, so you need a number for each stage:

• Picking. How many orders can one person pull per hour? Multiply by people and hours.
• Packing and staging. Same calculation, at the bench.
• Loading. How many orders fit in one vehicle, by volume and by weight, whichever runs out first.
• Driving. Stops per route per day. This is usually the real ceiling, and it’s the one people guess at instead of measuring.

For the driving number, don’t use an average speed and a map. Use your own delivery history: total stops completed divided by driver-days worked, over a month. If four drivers worked 20 days each and completed 4,800 stops, your operation runs at 60 stops per driver-day. That’s your baseline, and it already includes your real traffic, your real doorways, and your real paperwork.

Then track utilization, meaning what share of that capacity you’re actually using. Fleet and route utilization benchmarks published by delivery software vendors tend to land in the high 70s to mid 80s as a percentage, and the logic behind that range holds up regardless of the exact number: below it, you’re paying fixed costs on vehicles and salaries that aren’t moving goods; above it, you have no absorption left for a sick driver, a breakdown, or an unusually big order.

A worked example. A florist with two vans, two drivers, and an average of 45 stops per driver-day has a delivery capacity of 90 orders per day. If Valentine’s week brings 140 orders on a Tuesday, no amount of route optimization closes a 50-order gap. That gap has to be closed weeks earlier, with a third vehicle, a second wave of routes, extended delivery windows, or outside delivery capacity.

Finding the Bottleneck That Actually Limits You

Your operation has exactly one binding constraint at any moment, and everything you spend on the other stages is wasted until you fix that one.

The test is simple: walk the order flow and look for where work piles up. Orders waiting on the pack bench mean picking is outrunning packing. Packed orders sitting on the floor past van departure means loading or driving is the constraint. Drivers standing around at 8 a.m. means picking is the constraint and they’re waiting on product.

The trap is that the pile is a symptom, not always the cause. Orders stacking up at the pack bench might be a packing problem, or it might be that picking delivers them in a random order and packers spend their time sorting. That’s a layout problem wearing a packing problem’s clothes, and it gets solved in the building rather than at the bench, which is why how the shelves and picking paths are arranged belongs in any serious capacity conversation.

Once you’ve fixed the binding constraint, the constraint moves somewhere else. That isn’t failure; that’s how the process works. Fix picking and driving becomes the ceiling. Fix driving and your inbound supply becomes the ceiling. The job is to always know which stage currently governs.

Forecasting Demand Well Enough to Plan Against

You don’t need a statistics degree. You need last year’s numbers and a calendar.

Three inputs cover most of it:

• Your own order history, by week. Twelve to twenty-four months of weekly order counts shows both the trend and the shape of your year. Most delivery businesses have a seasonal shape far stronger than they assume.
• Known events. Holidays, local festivals, a wholesale customer’s promotion, the university term starting. These are not forecasts; they’re facts on a calendar, and they’re where capacity breaks.
• Committed pipeline. Standing orders and signed accounts are demand you already know about. Count them separately from speculative growth.

Add them, then plan against the peak week rather than the average week. Averages hide the days that break you. A business that averages 70 orders a day but hits 140 twice a year has a 140-order problem, not a 70-order operation.

Getting those three inputs into one place is usually the hard part, because order history sits in one system, standing accounts in another, and the calendar in someone’s head. Larger operations solve this by consolidating raw operational data into a single store first, the approach behind data lake consulting services. A single spreadsheet with a row per week does the same job at small scale, provided everything actually lands in it.

Forecast accuracy matters less than forecast timing. A rough number eight weeks out is worth more than a precise number on the day, because eight weeks is long enough to hire, rent a vehicle, or line up extra delivery support. On the day, you have only triage.

Lead, Lag, and Match: Three Ways to Add Capacity

There are only three timing strategies, and each is a different bet.

Most small operations lag by default, because lagging requires no decision. That’s the expensive path in a delivery business specifically: when you miss delivery windows, customers don’t wait in a queue, they go elsewhere. Research on delivery failure consistently finds that a large majority of shoppers won’t order again from a business after a bad delivery experience.

Match is the realistic middle for most businesses that deliver. The increments are things like a rented van for a peak month, a part-time driver for Fridays, or outside delivery capacity for the two weeks around a holiday. All of them are small commitments that can be reversed if the demand doesn’t hold.

Short-Term and Long-Term Capacity Decisions Are Different Jobs

Treat them separately or they’ll blur into permanent firefighting.

Short-term (this week to this quarter) is about using the capacity you already have. Resequencing routes, shifting delivery days for flexible customers, batching a neighborhood into one run, adding a Saturday, staggering start times so picking finishes before drivers arrive. None of this costs capital. All of it can add 10-20% of effective throughput in an operation that has never tuned it.

Long-term (six months to two years) is about changing what you have. Hiring drivers, buying or leasing vehicles, taking more warehouse space, signing on outside delivery partners, investing in systems. These decisions have lead times measured in weeks or months, and they’re the reason forecasting eight weeks out matters.

The common failure is using short-term tools on a long-term problem. If you’re resequencing routes every single morning to squeeze the day in, you don’t have a routing problem. You’re structurally under capacity, and no amount of clever scheduling fixes that.

Building Slack Into the Plan on Purpose

A plan that only works when everything goes right isn’t a plan.

Running at 100% of measured capacity means the first van breakdown, sick driver, or late supplier delivery turns into missed customer commitments. Deliberate slack is what absorbs normal variation: a buffer of unused capacity, a relief driver, a standing arrangement with an outside delivery provider.

Where that slack lives matters. Slack at a non-constraint stage is wasted money. Slack at the constraint is insurance. If driving is your ceiling, a spare pack bench helps nobody; a backup driver arrangement is what you should be paying for.

The same logic applies upstream. Your capacity plan assumes product shows up when expected, and when a supplier misses, your delivery capacity is irrelevant because there’s nothing to load. Keeping the inbound side steady is its own discipline, covered in our guide to building supply chain resilience.

Metrics Worth Tracking Every Week

Capacity planning is a loop, not a project. Four numbers keep the loop honest:

• Orders per driver-day. Your capacity baseline. If it drifts down, something in the operation got slower.
• Capacity utilization. Actual output divided by measured capacity. Watch the trend more than the level.
• On-time delivery rate. The first place capacity strain shows up, usually before anyone admits there’s a problem.
• First-attempt success rate. Failed deliveries consume capacity twice, once for the failed run and once for the redelivery, and industry estimates for the cost of a failed first attempt in the US cluster around $17 per parcel. A rising failure rate quietly eats capacity you thought you had.

Review them weekly, and re-measure your true stage-by-stage capacity twice a year. Operations change faster than people expect, and a capacity number from 18 months ago describes a business that no longer exists.

Where Capacity Planning Actually Pays Off

The return isn’t in a dashboard. It’s in the decisions you stop guessing at.

Whether to take on a wholesale account that adds 30 stops a week. Whether the answer to a busy December is a temporary van or a permanent hire. Whether the thing slowing you down is the driving or the 40 minutes every morning spent finding product in the back. Each of those is answerable once you know your real throughput, your real constraint, and your real forecast.

Start small. Measure one week of stops per driver-day, pull twelve months of order history into a weekly chart, and walk the floor once looking for the pile. Those three things take an afternoon and will tell you more about your capacity than a year of running on instinct.