Sales forecasting for delivery operations is the practice of estimating how much you will sell in a coming period, and then reading that estimate as a workload: how many orders go out, on how many routes, needing how many driver hours and how much stock on the bench.
That second half is what makes it different from forecasting in a software company or a showroom. When a wholesale bakery forecasts an extra 400 cases for the second week of December, the number is not a revenue target. It is two more routes, one more driver, a flour order placed eleven days early, and a cold storage problem if any of those three are wrong.
Most of the pressure sits on the delivery side of the business, because that is where the money goes. Last-mile delivery now accounts for around 53% of total shipping costs, up from roughly 41% in 2018 (SmartRoutes), and labour is 50% to 60% of last-mile cost on its own (Locus). Driver hours are the single largest line you commit in advance and the hardest to unwind once committed. A forecast is how you decide that number before the week starts instead of after.
This post covers what a forecast is for and what it changes about how you run the week. If you already accept the argument and want the mechanics, skip ahead to the step-by-step guide to building a sales forecast you can plan routes against.
The Bottom Line
- A sales forecast in a delivery business is a workload estimate, not a revenue goal. Its real output is routes, driver hours and stock quantities.
- Forecasting badly is expensive in both directions: 55% of small and midsize businesses carry at least 20% excess stock, and inventory management and overstocking were named by 51% of leaders as major contributors to waste (Avery Dennison, via FoodBev).
- Accuracy expectations should be modest. The median sales team lands at 70% to 79% accuracy and only 7% reach 90% or better (Gangly). A forecast that is directionally right and reviewed weekly beats a precise one reviewed annually.
- You do not need software to start. 62% of US small and midsize businesses still forecast in spreadsheets or legacy accounting systems (CFO Research and AICPA 2025 Finance Technology Study), and for a single-depot operation that is often the correct tool.
Boost customer satisfaction with just a few clicks
Most-Loved Features:
- On-demand drivers
- Real-time GPS tracking
- Delivery confirmation photos
- Over 50% of customers report a smoother delivery experience
What sales forecasting means in a delivery operation
A sales forecast is a written estimate of future sales volume for a defined period, built from order history, known commitments and whatever you can see coming. In a business that delivers its own orders, you build it in units before you build it in money, because units are what the operation consumes.
The practical form is a grid: period down one side, product or customer group across the top, and a quantity in each cell. A florist forecasting February wants stems and arrangements by week. A wholesale distributor wants cases by account. A caterer wants covers by event date. Revenue is derived from that grid afterwards by multiplying through the price list, which is the opposite of how most sales forecasting guides written for software companies approach it.
Three things distinguish it from general-purpose revenue forecasting:
- The unit of time is short. Quarterly forecasts are nearly useless for scheduling drivers. Weeks, and often days within a week, are the working granularity.
- Geography is part of the forecast. Forty extra orders clustered in one neighbourhood is one extra route. Forty spread across the metro is three.
- The forecast expires. A forecast for a perishable product has a shelf life measured against the product’s own shelf life. Being right about March in April does not help.
Demand forecasting vs sales forecasting: what is the difference?
Sales forecasting predicts what you will sell. Demand forecasting predicts what the market wants, whether or not you can supply it. The distinction sounds academic until you hit a capacity ceiling.
If your kitchen can produce 600 cases a week and you are selling 600 cases a week, your sales forecast says 600 and your demand forecast might say 800. The sales forecast is correct and useless; the demand forecast is the one telling you to add a shift or a vehicle. Small operators tend to forecast sales only, which means a business that has been capacity-constrained for a year has no record of how much it turned away.
The simple fix is a second column in the grid for orders you declined, cut short or pushed to a later date. It costs nothing and it is the only evidence you will ever have for an expansion decision.
What a sales forecast decides when you run delivery routes
The reason to forecast is that several decisions have to be made before the sales happen, and each has a lead time longer than the sales cycle.
Driver and vehicle commitments. Scheduling is the forecast’s most immediate customer. Staff to last week and a peak leaves orders sitting; staff to the peak every week and you pay for idle hours in the troughs. Because labour is the dominant last-mile cost, this is usually where forecast error shows up first in the margin.
Route structure. The number and shape of routes follows from volume and its geographic spread. Knowing on Thursday that next Tuesday is heavy lets you split a route in advance rather than discovering it at 6am when a van is already loaded.
Stock and ingredient buying. Purchase lead times frequently exceed customer order lead times. If flour arrives in seven days and customers order three days out, you are always buying against a forecast, whether you have written one down or not.
Cash. You pay drivers and suppliers this week and get paid on day 30 or 45. A volume forecast converted into a cash view is what tells you which week gets tight. That is a related but separate exercise, and it only becomes possible once the volume forecast exists.
Pricing and promotion. Knowing a slow week is coming is what makes a promotion a deliberate act rather than a panic.
The cost of running a delivery operation without a forecast
Forecast error is expensive in both directions, and the two failures look nothing alike.
Over-forecasting shows up as stock you cannot sell and hours you did not need. For anyone moving perishable goods the number is unforgiving: food waste is projected to cost the global retail supply chain $540 billion in 2026, up from $526 billion in 2025, and the report puts inventory management and overstocking among the main contributors, named by 51% of the leaders surveyed (Avery Dennison, via FoodBev). The same research found 55% of small and midsize businesses carrying at least 20% excess stock, and identified meat as the hardest category to manage at 50% of respondents, followed by produce at 45% and baked goods at 28%.
Under-forecasting shows up as service failure, which is slower to appear on a P&L and more damaging. Orders go out late, a route runs over, a standing wholesale account gets a short delivery twice in a month and starts taking calls from a competitor. The cost is the account, not the box.
The asymmetry matters for how you set the forecast. For a product with a two-day shelf life and a replaceable customer, lean low. For a standing contract with a restaurant group that expects full delivery every Tuesday, lean high and eat the waste. Most operators have both in the same book and need both rules.
How accurate does a sales forecast need to be?
Less accurate than people assume, and more consistent than people manage.
The benchmark commonly used in sales organisations is forecast accuracy within 15% (A Sales Growth Company), and most teams do not hit it: 79% of sales organisations miss their quarterly forecast by more than 10%, while the median team sits at 70% to 79% accuracy and only 7% clear 90% (Gangly). If professional sales teams with dedicated analysts land there, a bakery owner forecasting on a Sunday evening should not be aiming for perfection.
What does move accuracy is frequency. The same research found teams tracking weekly reached 87% accuracy against 52% for those tracking irregularly. A rough forecast you revisit every Thursday will beat a careful one you write in January and never open again, because the revisiting is where you learn the shape of your own demand.
A useful standard for a small operation: your forecast should be good enough that you are never surprised by the size of a week, and that you can explain any miss larger than about 15% after the fact. If you cannot explain the miss, you are missing a variable that belongs in the forecast.
Choosing a forecasting method for your delivery business
There are several ways to produce the number, and they differ mainly in how much history and how much admin they need. Historical and moving-average approaches extrapolate from what you have already sold and suit businesses with steady repeat orders. Pipeline approaches work from named opportunities and suit businesses closing a small number of large accounts. Bottom-up builds from individual customers or routes. Qualitative forecasting is structured judgement, and it is the only option in a first year of trading.
Most delivery businesses end up blending two: a historical baseline for the standing order book, plus judgement for events, weather and anything new. Which blend fits depends on your order mix and size, and the full comparison of sales forecasting methods for an operation that delivers its own orders sets out what each one needs and where each one breaks.
How to start forecasting if you have never done it
You need twelve months of order history if you have it, a spreadsheet, and about two hours. Pull orders by week, average the comparable weeks, adjust for what you know about the period ahead, and write the result down somewhere you will look at it again. Then compare it to reality at the end of each week.
The first forecast will be wrong. Its purpose is to exist, so the second one has something to improve on. The mechanics linked above go further, including how to pick a horizon and how to convert a unit forecast into routes and driver hours.
When forecasting software is worth paying for
For a long time, it is not. A single-depot business with a few hundred orders a week and a stable customer list can forecast adequately in a spreadsheet, and most do: only 38% of US small and midsize businesses use automated forecasting tools, leaving 62% on manual spreadsheets or legacy accounting systems (CFO Research and AICPA 2025 Finance Technology Study).
The case for buying something changes when the spreadsheet starts costing you time or trust. The usual triggers are multiple depots, more SKUs than one person can hold in their head, several people needing the same numbers at once, or a sense that nobody believes the file any more. That last one is real and measurable: 69% of US small and midsize business finance professionals said spreadsheet-based forecasting creates moderate to high operational risk (Trezy, 2026 survey). The categories of tool available and what to check before paying for one are set out in the review of sales forecasting software for businesses with delivery routes.
Seasonality and the delivery calendar
Seasonality is the part of demand you can see coming, and for most food, floral and catering operations it dominates the forecast. Valentine’s Day, Mother’s Day, Thanksgiving week and the fortnight before Christmas are not variance; they are the shape of the year.
Two practical notes. First, forecast peaks off the same week last year rather than off a recent average, because an average built from ordinary weeks will always under-call a peak. Second, forecast the days around the peak as carefully as the peak itself. The week after a holiday is frequently the quietest of the quarter, and operators who staffed up for the peak and forgot to staff back down pay for it twice.
Weather belongs in the same category for anyone delivering perishables or serving outdoor events. It is not predictable months out, but it is predictable five days out, which is inside the window where you can still move a driver shift.
What to do with the forecast once you have one
A forecast that lives in a file nobody opens is a writing exercise. Three habits make it operational:
- Compare forecast to actual every week, in writing. One column, two numbers, a note on why they differ. This is the entire feedback loop.
- Give the forecast to whoever schedules drivers. If the person building routes has not seen it, it has not affected anything.
- Keep the misses. A year of recorded misses tells you which weeks you systematically over-call, which is worth more than any method you could adopt.
Keeping that record has a second use beyond the operation. A forecast with a documented history of how it was built and where it missed is the same evidence base you draw on when an outside party asks for your numbers, whether that is a lender, an insurer, or an obligation in the list of reporting requirements a growing business has to meet. Businesses that forecast casually and never write anything down end up reconstructing the reasoning afterwards, which is considerably harder than keeping it as you go.
Frequently asked questions
What is sales forecasting in simple terms?
It is an estimate, written down in advance, of how much you will sell in a coming period. In a delivery business the estimate is kept in units such as orders, cases or covers, because those convert directly into routes, driver hours and stock quantities.
How far ahead should a delivery business forecast?
Run two horizons. A rolling four to six weeks in weekly buckets drives scheduling and purchasing, and a twelve-month view in monthly buckets drives hiring, vehicles and seasonal planning. Longer than twelve months is planning, not forecasting.
How much data do you need to start forecasting?
Twelve months of order history lets you see seasonality and is the comfortable starting point. With three to six months you can still forecast short horizons from recent trend. With none, you forecast from judgement and capacity, and you tighten it as the first months of real data arrive.
Is sales forecasting the same as a sales target?
No, and conflating them is the most common failure. A forecast is your honest estimate of what will happen. A target is what you want to happen. When the target gets written into the forecast, you staff and buy for a number nobody expects to hit.
Can you forecast accurately with a spreadsheet?
Yes, for most small operations. Accuracy comes from reviewing the forecast often and recording your misses, not from the tool. Software earns its place when the volume of data or the number of people needing it exceeds what one file can carry.
