Before AI, Connect Your OEE Data: Advice from Schneider Electric

On 2 October 2026, IIoT World published an article by Lucian Fogoros based on a video interview with Neil Smith, Segment President for Consumer Packaged Goods at Schneider Electric. The message is blunt: before talking about artificial intelligence, plants should fix their data foundation, starting with OEE and downtime reason codes. It is worth noting that the article is a short summary of an interview, written from a vendor's perspective; the page also states that AI tools helped summarize the content, with editorial review.

The problem: data that exists but doesn't talk

According to the interview, in many plants OEE is still captured on paper sheets, disconnected spreadsheets or non-integrated software. When a line stops, the reason code is assigned manually by the operator. The data is there, but the systems were not designed to share it, so it is hard to bring out the patterns behind recurring problems.

The article cites one figure: in US CPG manufacturing, delays and machine downtime would account for 18% of total product cost. The figure refers to the United States and the page does not give its source, so it should be read as a stated order of magnitude, not a verified benchmark, and it cannot be transferred as-is to the Italian context.

The priority: automate and connect

For Smith, the first step is to automate the assignment of reason codes and connect OEE systems to the plant network. Only when data flows can data science look for correlations. The example given is a recurring jam at the carton infeed, which might turn out to be linked to humidity in the area, outside temperature or work shifts.

Smith describes advanced process control (APC) as the "first generation of AI" and argues that the more sophisticated models rest on the same connected data infrastructure.

Closing the loop: from recommendation to action

An AI recommendation, such as changing a setpoint or restarting a line, has to turn into action on the shop floor. According to Smith, legacy control systems have no native path from enterprise AI platforms to the machines. He proposes open, software-defined automation as an "action broker" between the two layers, keeping existing controls in place and migrating gradually. This part reflects Schneider Electric's product vision and should be assessed as such.

The interview also points to operators' knowledge: they know the plant better than anyone, and capturing it becomes more valuable as the workforce ages and hiring gets harder.

Why it matters for Italian plants

The article contains no case studies or measured results, and it concerns CPG and the US market, not Italy or the poultry sector. Yet the logic is general and applies equally to food and poultry processing lines: without automatic, consistent line data, any advanced analytics starts from a fragile base. This is exactly the role played by MES and production monitoring systems.

What to do in practice

A few practical steps, consistent with the approach of the interview:

• Review how OEE and downtime are captured today: paper, spreadsheets or integrated systems.
• Reduce manual entry of reason codes by acquiring machine states directly from the line.
• Connect systems to the plant network so that downtime data can be compared with other variables (shifts, environmental conditions, product).
• Involve operators in defining reason codes, so their experience is not lost.
• Only then, evaluate advanced analytics or AI on reliable data.

Connected, automatic data first, then artificial intelligence.