A dashboard nobody opens is a common and expensive outcome. It usually happens because the dashboard answers a question that was interesting during the project and not one anybody needs on a Monday morning. The articles below start from the decision and work backwards to the data.

The practical ground: getting figures out of an ERP without destabilising it, reconciling numbers that two systems report differently, choosing between a reporting replica and a warehouse, and knowing when a well-built export is genuinely the right answer instead of a platform.

Forecasting and anomaly detection appear where they earn their place — stock-outs, late payment, unusual consumption — and not as a way of putting machine learning on a slide.

41 articles in this topic AI & business intelligence solutions