Outlier removal: clean noise from your data without ruining your forecast

If you work with demand forecasting, this will sound familiar: you have what looks like a solid history, you run your model… and the output still doesn’t stack up. That’s why we’ve put together our downloadable guide on outlier removal in forecasting.

Because, very often, the issue isn’t the model. It’s the data. More specifically, it’s the outliers.

These unusual values (demand spikes, one-off errors or abnormal behaviours) may look like isolated cases, but their impact is far bigger than it seems. In fact, as this resource explains, a single value that’s out of place can pull the whole model off course.

This document is designed to help you tackle that problem in a practical way, without unnecessary complexity.

You’ll see why removing outliers is not as simple as applying a standard formula. Many companies still use classic methods such as Z-score or IQR directly on historical data, but in real-life environments (with seasonality, campaigns or trends) this often creates more problems than it solves.

Why? Those methods aren’t built for time series. Applied without context, they can flag “anomalies” that are actually completely normal for the business.

This document shows exactly that: why traditional approaches fail and what to do instead.

Instead of working on the series directly, we recommend a far more robust approach: detecting outliers on the model residuals.

It might sound technical, but it’s explained clearly and with a practical focus. In simple terms, you first isolate the expected behaviour (trend, seasonality) and then identify which values truly deviate from what should have happened.

That shift in approach is critical. The goal isn’t just to “clean data”. It’s to do it without losing valuable information or distorting the series.

What you’ll find in this document

• What outliers are in forecasting (beyond the textbook definition).
• Why classic methods don’t work well for time series.
• Real examples of common mistakes in anomaly detection.
• A practical residual-based approach to detect outliers properly.
• Key principles to improve data quality and increase forecast accuracy.

Download it now and start looking at your forecasts with a new level of precision.