---
title: "Demand Forecasting with AI: Tools, Data, Accuracy, Cost"
url: https://cruxdigits.nl/vraagvoorspelling-met-ai/
language: en
description: "Demand forecasting with AI: what it adds to Excel and your ERP, the data you need, MAPE and bias explained, and what it costs. Fixed price per step."
publisher: Crux Digits B.V. (https://cruxdigits.nl)
---

Demand forecasting and planning

# Demand forecasting with AI: what it delivers and what it costs

AI demand forecasting earns its place where Excel and your ERP's forecast module only extend your sales history. A model that also weighs promotions, weather and holidays forecasts each item more sharply and shows how certain each forecast is.

What AI adds to Excel and your ERP, the data you need, and how to measure the gain on your own figures.

[Free consultation](https://cruxdigits.nl/#contact) [Pricing](https://cruxdigits.nl/pricing/)

Share

In short

**A better demand forecast pays back on two sides: you miss fewer sales and you waste less stock, capacity and staff time.** For a footwear manufacturer with €235M in turnover, Crux Digits built a forecasting and planning layer on Prophet and XGBoost; within a year revenue grew 8% and production capacity 7% ([read the case](https://cruxdigits.nl/case-studies/demand-forecasting/)). A first test on your own data costs €2,500, a working forecast for one product group €20,000 ([all prices](https://cruxdigits.nl/pricing/)).

Three fixed facts

## What is settled before the model starts

- 2 yearsSales history as the starting pointPer item, location and week, with stock-out days flagged. Prophet, one of the models we use, works best with several seasons of history.
- €2,500AI audit in 1 to 2 weeksThe MAPE and bias of your current method measured on your own data, with a fixed price for the first build.
- 4 to 6 weeksProduction-ready MVP, €20,000One product group, backtested against your current method and run beside the planner for a few weeks.

Sources: [Prophet on CRAN](https://cran.r-project.org/web/packages/prophet/index.html); [Hyndman and Athanasopoulos on MAPE](https://otexts.com/fpp3/accuracy.html); our [pricing](https://cruxdigits.nl/pricing/).

From history to adjustment

## How a demand forecast comes about

Five steps from sales history to a plan that purchasing, production and rostering can build on, and what the planner sees at each point.

1. History ### The sales history Per item, per location and per week, with stock-out days flagged: an item that was out of stock sold nothing, but the demand was there.
2. Signals ### Promotions, weather and holidays What shifts demand is not in the history: the promotion calendar, the KNMI weather forecast, school breaks and seasons.
3. Model ### The model learns and is backtested Season and trend from the history, the drivers on top. Tested on held-out periods against your current method, on MAPE and bias.
4. Planning ### A proposal in the plan The forecast becomes a reorder point, a production run or a roster, as a proposal in the package where purchasing and planning already work.
5. Adjust ### The planner keeps the final say The planner adjusts or overrides. That correction becomes data for the next round, and the MAPE is measured again every month.

What it delivers

## What a better demand forecast delivers

A demand forecast is not a report. It is the number purchasing, production and rostering base their decisions on. When that number runs high, cash sits in slow stock. When it runs low, you turn customers away on exactly the items they want, or pay overtime to catch up.

What you recover depends on where the error lands today: less stock at the same service level, fewer changeovers on the production line, or rosters that match the workload. At the footwear manufacturer in our [demand forecasting case](https://cruxdigits.nl/case-studies/demand-forecasting/), most of the capacity gain came from mould changes that now lined up with real demand instead of reacting to surprises. Those are the figures of one engagement, not a promise for yours. That is why we first measure how well you forecast today, so the gain becomes a number on your own data.

Excel, ERP or AI

## What AI adds to Excel and your ERP's forecast module

| Approach | What it does | Where it stops |
| --- | --- | --- |
| Excel | FORECAST.ETS uses the AAA version of exponential smoothing and detects seasonality automatically ([Microsoft](https://support.microsoft.com/en-us/office/forecast-ets-function-15389b8b-677e-4fbd-bd95-21d464333f41)). | One series at a time, no promotions or weather, and upkeep sits with one planner. |
| Business Central | The Sales and Inventory Forecast extension predicts sales with Azure AI from your sales history and suggests replenishment ([Microsoft Learn](https://learn.microsoft.com/en-us/dynamics365/business-central/ui-extensions-sales-forecast)). | One aggregated forecast for all locations and variants; no prediction when data is too thin or too variable. |
| Odoo | The Master Production Schedule sets forecasted demand against stock and production ([Odoo](https://www.odoo.com/documentation/18.0/applications/inventory_and_mrp/manufacturing/workflows/use_mps.html)). | The user enters the forecasted demand; the package does not calculate it. |
| SAP IBP | Demand sensing creates daily forecasts for the next 4–8 weeks, starting from the existing demand plan ([SAP Learning](https://learning.sap.com/courses/mastering-sap-ibp-for-demand/introduction-to-demand-sensing_e8fa54aa-2e8c-4a0d-8bf6-2055b09e8669)). | It sharpens the short term and does not replace demand planning itself. |
| Your own AI model | Learns from history plus promotions, prices, weather and holidays, per item and per location, and is tested against your current method. | Needs clean data and someone to review the output. |

What you get

## Three parts, each with a checkpoint for the planner

- ### A forecast per item and location History plus promotions, weather and holidays, with a range around every forecast, backtested against your current method. [Machine learning on your own data](https://cruxdigits.nl/services/machine-learning/)
- ### A proposal in your ERP or planning tool Reorder points, production runs or a roster come back as a proposal in Business Central, AFAS, Exact or your WMS. The planner approves. [What AI delivers for a wholesaler](https://cruxdigits.nl/wat-levert-ai-een-groothandel-op/)
- ### Measured on your own figures The MAPE and bias of your current method set the baseline in the audit, so the model's gain is a number before you invest further. [How the AI audit works](https://cruxdigits.nl/services/ai-audit-strategy/)

Data

## The data you need

- **Sales history at the level you plan.** Per item, per location and per week or day. Prophet, one of the models we use, works best with strong seasonal effects and several seasons of history ([CRAN](https://cran.r-project.org/web/packages/prophet/index.html)); two years is a reasonable start.
- **Stock-out days.** An item that was out of stock sold nothing, but the demand was there. Without that flag a model learns the wrong lesson.
- **Promotions and prices.** When, which item, what discount, through which channel. This is often scattered across the ERP, the webshop and a marketing calendar.
- **Weather.** In the Netherlands the KNMI publishes daily data per station, including temperature, sunshine and precipitation ([KNMI](https://www.knmi.nl/nederland-nu/klimatologie/daggegevens)).
- **Holidays and school breaks.** Prophet models public holiday effects out of the box; school holidays and local events you add yourself.
- **One condition.** A driver only helps if you know its future value when you forecast. A weather forecast reaches a few days ahead, a promotion calendar months.

Accuracy

## MAPE and bias: how to tell whether a forecast is any good

**MAPE** is the mean absolute percentage error. If you sold 100 units and the model forecast 90, the error is 10%; averaged over all weeks, that is the MAPE. It is easy to explain, with one weakness: when actual sales are zero it is undefined, and for items that rarely sell it swings wildly ([Hyndman and Athanasopoulos](https://otexts.com/fpp3/accuracy.html)). For slow movers, measure total absolute error divided by total sales instead.

**Bias** is the average error with its sign kept: do you forecast too high or too low, consistently? A model with a tidy MAPE and a bias of a few percent upwards quietly builds stock every week. Steer on both.

The number that matters most is the difference against your current method, measured over the same periods. That is why we first backtest a model on held-out history, as in the case, before it touches a decision.

Use

## Demand forecasting in the supply chain: production, inventory and workforce planning

- **Production planning.** The forecast sets run sizes and sequence, so changeovers line up with real demand. At the footwear manufacturer this produced most of the capacity gain.
- **Inventory and purchasing.** Reorder points and safety stock follow from the forecast and from its uncertainty. An item with a wide range around its forecast gets more buffer than a stable fast mover.
- **Staffing.** A forecast per week and per site or shift becomes a roster that matches the workload, with fewer last-minute temps. In workforce management this is called labour forecasting: the same demand forecast, translated into hours needed per role and per shift.
- **The planner keeps the final say.** The system proposes and shows why; the planner adjusts or overrides, and that correction becomes data for the next round. The model supports the judgement of the people who know the floor rather than replacing it.

Buy or build

## Forecasting software or a custom model on your own data

| Your situation | Sensible choice | Why |
| --- | --- | --- |
| Few items, short history, one planner | Excel with fixed rules | A model cannot learn what is not there yet. Start with clean data and one way of measuring. |
| You run Business Central online and demand mostly follows history | The built-in extension first | It comes with your subscription. Measure its MAPE and see where it falls short. |
| A large planning team in SAP with an S&OP process | Your ERP vendor's planning software | The process, roles and integration already exist; custom work fills a gap at most. |
| Promotions, weather or locations drive demand, or production has hard constraints | A custom [machine learning](https://cruxdigits.nl/services/machine-learning/) model on your own data | Packaged tools do not weigh those drivers, or not per location; your own model does, and you own the code and the model. |

Who does what

## Who does what alongside your ICT partner

| Party | Responsible for | Not for |
| --- | --- | --- |
| Crux Digits | Data pipeline, model, backtest against your current method, forecasts as proposals in your ERP or planning tool, and handover of code and test set. | Your network, workplaces, hosting or helpdesk. |
| Your ICT partner | Identity and access, the environment the model runs in, monitoring and backup under your existing agreements. | The content of the forecast. |
| Your IT or operations manager | Ownership, priorities, the change through your change process and the link with the ERP partner. | Building the model. |
| Your ERP partner | Reviews the integration for the APIs and permissions used, and makes any change inside the package itself. | The model and the data pipeline. |
| Planners and buyers | Reviewing, adjusting and releasing the proposals. | Maintaining the pipeline. |

Cost and timeline

## What AI demand forecasting costs and how long it takes

| Step | What you get | Duration | Price excl. VAT |
| --- | --- | --- | --- |
| AI Audit & Strategy | Your sales, stock and promotion data reviewed, the MAPE and bias of your current method measured, and one scoped first build at a fixed price. | 1–2 weeks | €2,500 |
| Production-ready MVP | A forecast for one product group, site or line on your own data, backtested against your current method and run beside the planner for a few weeks. | 4–6 weeks | €20,000 |
| Production | All relevant groups, integrated with your ERP or WMS, with monitoring, documentation and handover to your IT organisation and ICT partner. | 8–12 weeks | from €50,000 |
| After delivery | Your IT organisation runs the model. If a small team would rather not, [managed AI support](https://cruxdigits.nl/ai-beheer-mkb/) covers monitoring, model updates and up to 4 hours of improvements a month. | Ongoing | €500 a month |

Examples

## Four examples by sector

- **Wholesale (example).** A forecast per item and customer group feeds reorder points in the ERP; buyers see a proposal with its uncertainty. More in [what AI delivers for a wholesaler](https://cruxdigits.nl/wat-levert-ai-een-groothandel-op/).
- **Food production (example).** Short shelf life makes every error expensive. Weather and holidays drive daily demand for fresh and chilled lines; the forecast sets the production plan per line, and the planner sees expected sales per product for the coming days.
- **Energy (example).** A company forecasts its own consumption per quarter hour against its contracted capacity and moves batches to quiet hours. See [grid congestion in industry](https://cruxdigits.nl/netcongestie-industrie/).
- **Retail (example).** A forecast per store and per week, with promotions and holidays, feeds both store replenishment and the staff roster.

FAQ

## Frequently asked questions

### What is demand forecasting?

Demand forecasting is estimating how much customers will buy of an item or service in a coming period, per item, location and period. Purchasing, production and rostering build their plans on that number. Demand forecasting with AI uses drivers such as promotions, prices, weather and holidays next to sales history, and states how certain each forecast is.

### How does AI demand forecasting work?

A model learns from your history which patterns recur, such as season, trend and holidays, and which drivers shift demand, such as a promotion or warm weather. In our case we combined Prophet for season and trend with XGBoost for the drivers. The model is first backtested on held-out periods and only then used for planning.

### How accurate is AI demand forecasting?

It depends on your items and your data, so there is no honest number in advance. What is certain: you measure it with MAPE and bias, over the same periods as your current method. We measure that baseline in the audit and backtest the model against it, so the improvement shows on your own figures before you invest further.

### What does custom demand forecasting software cost?

It starts with a €2,500 audit in 1–2 weeks. A Production-ready MVP for one product group costs €20,000 in 4–6 weeks; production starts from €50,000 in 8–12 weeks. All excl. VAT. If you go on to the MVP within 60 days, the audit is credited in full. Code, model and test set are handed over to you.

### How much historical data do you need?

For seasonal patterns a model needs several seasons, so two years of weekly sales history is a reasonable start. For new items with little history the model borrows from similar items or product families. Quality matters more than length: stock-out days flagged, promotions recorded, and item codes that mean the same thing across the years.

### What is seasonal demand forecasting?

A forecast that takes the recurring pattern in demand into account: the weekly cycle, the yearly season, public holidays and school breaks. Excel detects such a pattern with FORECAST.ETS, but for one series at a time. A model such as Prophet handles several seasonalities at once, per item and location, and layers promotions and weather on top. The condition is that your history spans several seasons; otherwise the model has nothing to learn from.

### Machine learning or statistical forecasting: what is the difference?

Statistical methods such as exponential smoothing and ARIMA extend one series from its trend and season; they are fast, explainable and often good enough for stable items. Machine learning, such as a gradient-boosting model, also learns from drivers like promotions, price, weather and holidays, and across items. In practice we combine the two: Prophet for season and trend, XGBoost for the drivers, and the backtest on your own data decides which approach wins per item group.

### Do we need to replace our ERP or planning system?

No. The forecast runs as its own layer beside your ERP and reads sales and stock through the integrations the package offers. The output returns as a proposal in the planner's or buyer's workflow, who reviews and releases it. Your ERP partner and ICT partner keep their own roles, and a package upgrade does not touch the model.

## Want to know what a better forecast is worth to you?

A free 30-minute call: you tell us how you forecast today and what data you have, we tell you whether an audit is worth it and what the first step costs.

[Book a free consultation →](https://cruxdigits.nl/#contact)
