---
title: "Inventory Forecasting: Methods and Examples | Ant Assist"
description: "Inventory forecasting for UK online sellers: the main methods, a step-by-step process, a worked example in Excel, how to check accuracy and where AI fits."
url: "https://ant-assist.com/resources/inventory-forecasting/"
---

RESOURCES · ECOMMERCE · INVENTORY FORECASTING

# How to do inventory forecasting

A moving average of recent sales, adjusted for seasons and promotions, is the simplest sound inventory forecast for a small seller. Cleaner data beats cleverer maths.

Tools our VAs use for this

- Updated 23 October 2026
- 13 min read
- By Ant Assist

[Calculate your savings](https://ant-assist.com/calculator/) · [Book a free 15-min call](https://ant-assist.com/book/)

In short

Inventory forecasting estimates how much of each product you will sell over a coming period, then turns that estimate into how much stock to hold and when to order it. For a small online seller the simplest sound method is a moving average of recent sales, adjusted for seasons and planned promotions, multiplied out over your supplier's lead time. Check each forecast against what actually sold and adjust the method when the error grows. If you would rather someone else ran the weekly numbers, our [inventory management support](https://ant-assist.com/services/ecommerce/inventory-management/) gives you a dedicated VA to do it.

On this page

E-commerce Virtual Assistant

[Book a free 15-minute discovery call](https://ant-assist.com/book/)

On this page

[Book a call](https://ant-assist.com/book/) · [Calculate your savings](https://ant-assist.com/calculator/)

Part 2

## What inventory forecasting is

### Inventory forecasting defined

Inventory forecasting is a reasoned estimate of the stock you will need, not a prediction you can rely on blindly. Inventory forecasting asks, line by line: how many units of this will I sell next week, next month or next season, and how much stock do I need to cover that without overbuying? The answer drives your purchase orders.

### The data it needs

You need three things for each product line (SKU): sales history, by day or week, for as long as you have it; the supplier's lead time, measured from your own purchase records; and anything that will make the coming period different from the past, such as seasons, promotions, price changes, new listings or a competitor going out of stock. Stock-outs in your history matter too: a week with no sales because you had nothing to sell is not a week of no demand. If keeping those three things up to date is the sticking point, an [e-commerce virtual assistant](https://ant-assist.com/services/ecommerce/) can take it on.

Part 3

## How it differs from demand forecasting and replenishment

### Demand forecast first, stock plan second

Demand forecasting estimates what customers will buy. Inventory forecasting takes that demand estimate and turns it into a stock plan: how many units to hold, allowing for what you already have, what is on order and how long the next delivery takes. Replenishment is the action at the end: placing the order. The three are often run together in one spreadsheet, but keeping them apart in your head helps you spot which one went wrong when stock runs short.

Three linked steps: what customers will buy, how much to hold, then the order.

### Where reorder points fit

A reorder point is the stock level at which you place the next order. The ACCA, a professional accountancy body, describes the re-order level as the demand during the lead time in its [inventory control article](https://www.accaglobal.com/gb/en/student/exam-support-resources/fundamentals-exams-study-resources/f2/technical-articles/stock-control.html). Your forecast supplies that demand figure. Our [reorder point formula](https://ant-assist.com/resources/reorder-point-formula/) guide covers the calculation and safety stock in detail.

Part 4

## Why forecasting matters

### Fewer stock-outs and overstock

Without a forecast, you reorder when you notice a line is low, which is often too late, or you buy big to be safe, which leaves you holding stock. A forecast puts the order in at the right time and in the right size.

### Cash tied up in stock

For most online sellers, stock is the largest use of cash. A forecast that is roughly right lets you buy less, more often, for the lines that sell steadily, and saves the larger orders for the lines and seasons that need them.

### Supplier order planning

Suppliers with long lead times, minimum order quantities or busy seasons of their own need notice. A forecast tells you what to order and when, so you can book production or shipping slots early instead of paying for rushed freight.

Part 5

## The main forecasting methods

The main inventory forecasting methods are qualitative methods based on judgement, time-series methods such as moving averages, and causal and trend methods.

### Qualitative methods

Qualitative methods rely on judgement: your own knowledge of the market, feedback from customers, a supplier's view of the trend or a marketing plan. They are the only option for a new product with no sales history and a useful check on any number-based forecast. Their weakness is optimism, so write down the reasoning and compare it with what happened.

### Time-series methods and moving averages

Time-series methods assume the recent past is the best guide to the near future. A simple moving average takes the last few periods of sales and averages them; a weighted moving average gives more weight to the most recent period. These methods are easy to run in a spreadsheet, easy to explain and a sound default for steady lines.

### Causal and trend methods

Trend methods fit a line through your sales history to see whether demand is rising or falling. Causal methods link sales to something that drives them, such as price, advertising spend or the weather. They can be more accurate for lines with a clear driver but need more data and more skill to set up, which is why they usually live inside forecasting software.

### Choosing a method

Start with a moving average for every steady line. Add a seasonal adjustment for lines with a clear yearly pattern, and use judgement for new products and one-off events. Move to trend or causal methods only for your most important lines, and only when the simple method keeps missing in the same direction. Use ABC analysis, covered in our guide to [stock control methods](https://ant-assist.com/resources/stock-control-methods/), to decide which lines deserve the extra effort.

Part 6

## How to forecast inventory step by step

### Step 1: Gather the data

Export sales by SKU by week from your sales platforms, combined if you sell on more than one channel. Add your current stock, what is already on order and each supplier's lead time from your purchase records. Mark weeks when a line was out of stock or on promotion.

### Step 2: Pick a method and a period

Choose a forecast period that matches how often you order: weekly for fast lines, monthly for most. Pick a moving average over enough periods to smooth out noise but short enough to reflect recent change. Use the same method across similar lines so the results can be compared.

### Step 3: Adjust for seasons and promotions

Look at the same period last year. If a line sold noticeably more in that month than its average, carry that lift into this year's forecast, scaled to this year's level. For planned promotions, estimate the lift from your own past promotions, not from hope. For new products, borrow the sales pattern of a similar existing line and review it weekly. Writing the page for a new line is covered in [how to write product descriptions with examples](https://ant-assist.com/resources/how-to-write-product-descriptions/).

### Step 4: Review and update

Compare the forecast with actual sales every period, record the error and update the forecast. Fast-moving or seasonal lines need a weekly look; steady lines can be reviewed monthly. Revisit the method itself each quarter, or when a line's error keeps growing.

Keep it current

Have the forecast refreshed every week

Your VA pulls sales from each channel, marks stock-outs and promotions, refreshes the averages and flags the lines that need an order.

[Explore inventory management](https://ant-assist.com/services/ecommerce/inventory-management/) · [See your savings](https://ant-assist.com/calculator/)

Part 7

## Formulas and a worked example

### Average daily sales and lead-time demand

The two inventory forecasting formulas you need are average daily sales, which is units sold divided by days in the period, and lead-time demand, which is average daily sales times the lead time in days.

- Average daily sales = total units sold in the period ÷ number of days in the period
- Forecast demand over the lead time = average daily sales × lead time in days

The second is the demand-in-lead-time figure the ACCA uses for the re-order level.

### Worked example with invented figures

This is an example, not data from any real business. An online shop sells a single product line. Its example sales for the last three months are:

Worked example with invented figures

| Example month | Units sold |
|---|---|
| **Month 1** | 600 |
| **Month 2** | 660 |
| **Month 3** | 720 |

Step 1, three-month moving average: (600 + 660 + 720) ÷ 3 = 1,980 ÷ 3 = 660 units a month.

Step 2, average daily sales, taking a 30-day month for this example: 660 ÷ 30 = 22 units a day.

Step 3, forecast demand over the supplier's lead time, which is 14 days in this example: 22 × 14 = 308 units. That is how many units the shop expects to sell between placing an order and receiving it.

Step 4, adjust for a promotion. In this example the shop plans a promotion next month and, from its own past promotions, expects sales 10% above normal: 660 × 1.10 = 726 units for the month, or 726 ÷ 30 = 24.2 units a day.

Example only: invented figures from the worked example above.

Notice that sales rose every month in this example, so the three-month average of 660 sits below the latest month of 720. That is the main weakness of a simple moving average: it lags a trend. If the next month comes in near 780, the forecast was low for a reason, and a weighted average or a trend method would serve this line better.

### Building the forecast in Excel

Put one SKU per row and one column per week or month of sales. A moving average column uses AVERAGE over the last few sales columns. Add columns for lead time in days, average daily sales (the moving average divided by the days in the period) and forecast demand over the lead time (average daily sales multiplied by lead time). Add a seasonal or promotion factor column, defaulting to 1, that multiplies the forecast. Keep last year's same-month sales next to this year's so the seasonal pattern is visible.

Part 8

## Measuring and improving forecast accuracy

### Forecast error measures

Forecast error is the gap between what you forecast and what you sold. Record it every period for every important line. To compare lines of different sizes, express the error as a share of actual sales: in the worked example, if the shop forecast 660 units and sold 600, the error was 60 units, or 60 ÷ 600 = 10% of actual sales. Watch two things: the size of the error, and whether it is always in the same direction. An error that keeps landing on the same side is bias, and it means the method or the inputs need changing.

Forecast error as a share of actual sales, from the guide's invented example.

### Common causes of error

The usual causes are simple: stock-out weeks counted as low demand, promotions left in the history as if they were normal, a lead time taken from the supplier's quote instead of your own records, sales from one channel missing from the data, or a trend the method lags behind. Most accuracy gains for a small seller come from cleaner data, not cleverer maths.

Part 9

## Forecasting software and AI

### What forecasting tools do

Forecasting tools pull sales history from your channels automatically, apply one or more methods to each SKU, adjust for seasonality and suggest purchase orders. Many sit inside [inventory or order management systems](https://ant-assist.com/resources/ecommerce-inventory-management/). They save time once your range is too large to forecast by hand, but they are only as good as the stock and sales data they read.

### Can AI or ChatGPT forecast inventory?

AI tools can help: they can clean up a sales export, suggest a method, write the spreadsheet formulas, spot unusual weeks and summarise trends in plain words. A general chat assistant does not know your stock-out weeks, your supplier's real lead time or next month's promotion unless you tell it, and it can produce a confident answer from bad inputs. Treat its output as a draft forecast to check, never as a purchase order.

### What to double-check

Before acting on any forecast, from software or AI, check that the sales history is complete across every channel, that stock-out and promotion weeks are marked, that lead times come from your own records, and that the result makes sense against last year. If the forecast tells you to order far more or far less than usual, find out why before you place the order.

Part 10

## Handing the forecast review to a virtual assistant

A forecast is only useful if someone updates it every week: pulling sales from each channel, marking stock-outs and promotions, refreshing the averages, recording the error and flagging lines that need an order. That is steady, rule-based work, and it suits a dedicated virtual assistant, with you making the buying decisions. See our list of [ecommerce tasks to delegate](https://ant-assist.com/resources/ecommerce-tasks-to-delegate/) for how it fits with the rest of the back office.

At Ant Assist your VA is one named person, never shared, working full time from our own managed office in Cairo, where every VA is recruited, managed and supervised. They are not freelancers and not AI, and the service is UK-managed. Your VA works UK business hours, 9am to 5pm Monday to Friday, from 20 hours a week up to a full 40-hour VA, and if the fit is wrong we replace the VA without restart cost. Access sits under NDAs and data-processing agreements, runs through password managers and is revocable.

The rate starts at £5 per hour for the first 6 months on the Entry tier (3 months on Mid, 1 month on Expert); after that, Entry is £6.50 per hour full time (£1,040 a month for 160 hours) or £6.80 per hour part time (£544 a month for 80 hours). Try your own hours in the [calculator](https://ant-assist.com/calculator/). If you want help beyond stock, [outsourced e-commerce support](https://ant-assist.com/services/ecommerce/) covers the rest of the store.

Rated 5.0 on Google. · Trusted by our clients

FAQ

## Questions about forecasting stock

### What is short-term demand forecasting?

Short-term forecasting looks a few weeks to a few months ahead, which is the range most small sellers order within. It leans on recent sales and known events such as promotions, so it is usually easier to get right than a long-range forecast. Longer forecasts are still useful for planning big seasonal orders, but they need checking more often.

### What are the types of forecasting?

Guides group them differently, so no single list is official. The common split is judgement-based methods, time-series methods such as moving averages, trend methods and causal methods that link sales to a driver like price or advertising. Most small sellers need only the first two, with trend methods for their most important lines.

### Can inventory forecasting be automated?

Yes, much of it can. Software can pull sales data, run the methods and suggest orders on a schedule. A person still needs to mark promotions and stock-outs, check the output against common sense and make the buying decision.

### How do you forecast sales for new products?

With no history of its own, borrow the pattern of a similar line you already sell and scale it to your expectations. Start with a cautious first order, check sales every week and update the forecast as soon as real data comes in. Supplier and customer feedback help fill the gap early on.

Didn't find your answer? Browse [more common questions](https://ant-assist.com/help/) in the Help Center.

Get Started

Order stock before you run out.

One dedicated ecommerce VA, from £5 per hour. Book a call.

[Book a free 15-min call](https://ant-assist.com/book/) · [Calculate your savings](https://ant-assist.com/calculator/)

Discovery call · Role brief · VA matched · Onboarded into your tools