Blog

Your plant hit its targets, so where did the loss go?

The shift ran. The line hit its targets. The daily production report looked exactly as expected. But somewhere inside that shift, a changeover took twelve minutes longer than it should have.

It did the same thing yesterday. And the day before that. Not because equipment failed or someone made a mistake, but because a step in the changeover procedure no longer served a purpose. The equipment it was written for had been replaced years earlier, but the procedure never changed.

Nobody questioned it because nothing appeared wrong. It didn’t trigger downtime alarms or impact OEE enough to draw attention. It passed audits. It simply became “normal.” But twelve extra minutes per changeover becomes hundreds of hours over a year.

At The Marzetti Company, that added up to nearly $1 million in avoidable cost.

This is how manufacturing losses hide. Not because the data is missing, but because the context never makes it into the report.

The problem isn’t data

Most food and beverage manufacturers already have plenty of data. They track production, quality, downtime, OEE, yield, and energy consumption. A shift can hit every target while still containing:

• Excessive giveaway
• Extended sanitation cycles
• Process deviations
• Energy spikes
• Recurring micro-stops

The KPI stays green while profit quietly leaks away. When teams need answers, the process is familiar: export historian data, pull spreadsheets, align timestamps, and spend hours rebuilding what happened.

The data exists. The context is what’s missing.

What teams actually need to know

When engineers investigate a problem, they’re usually trying to answer a few simple questions:

• What happened?
• What was different from a normal run?
• What was happening before and after the event?
• Has it happened before?
• Is it an isolated issue, or part of a larger pattern?

If those answers require a full day of spreadsheet work, hidden losses often remain hidden because they rarely look dramatic.

They look like:

• Changeovers that consistently take longer than standard
• Yield drift after formulation changes
• More filler priming on night shift than day shift
• Quality holds tied to upstream conditions nobody connected
• Utility consumption that gradually trends upward

Individually, these seem minor. Collectively, they can cost hundreds of thousands – or even millions – of dollars per year.
 

How some plants find these losses faster

The difference isn’t better people. It’s the ability to investigate operational events in context.

Instead of manually stitching together historian data and spreadsheets, teams can compare good runs to bad runs, identify patterns across shifts and lines, and reuse analyses rather than rebuilding them every time.

The goal evolves from simply identifying that something changed to understanding what changed, why it changed, and how to prevent it from happening again.

The Seeq platform is built around this investigative workflow rather than static reporting. In practice, that involves:

• Searching and organizing signals across multiple data sources and systems without manual exports
• Comparing operating periods side by side to isolate what changed
• Overlaying batches or production runs on a common timeline to spot deviations quickly
• Identifying multivariate relationships that explain, rather than describe, a process condition
• Creating reusable analyses that don’t have to be rebuilt from scratch for the next investigation
• Monitoring ongoing performance so a resolved issue can be tracked, not just closed out

The value here is in the workflow, not any single feature: an investigation that used to require pulling data from three systems and reconciling it manually can instead start from where the process needs to be understood.

Put simply: the insight is only as valuable as the organization’s ability to act on it consistently, not just once.

What manufacturers have found

Marzetti identified a recurring idle period during production transitions that eventually amounted to nearly $1 million in avoidable cost.


Chobani saved an estimated 250,000 to 300,000 pounds of yogurt base in a single year by uncovering patterns in yogurt filler priming events.


Eli Lilly reduced a review process covering three years of manufacturing history, 21 units, and 3,000+ batches from more than two days to roughly two hours.


Different companies. Different processes. The same pattern: Organize the data > Put events in context > Compare what happened > Find the loss.

What could be hidden in your existing data?

Most manufacturers already have years of production, process, quality, and equipment data. The question isn’t whether the data exists, but whether your team can use the data to quickly answer:

• Which changeovers consistently exceed standard time?
• Which process deviations never trigger formal investigations?
• Which quality issues share the same root causes?
• Which recurring losses have simply become accepted as normal?

The answers are probably already in your historian; they just need the right context.

Sometimes the biggest losses in a plant aren’t the ones that show up in the report. They’re the ones the report was never designed to reveal.

Explore the recent Marzetti webinar and see how contextual analytics helped uncover an operational loss that traditional reporting had missed.


Skip ahead to the section that aligns most with your interests:

Seeq is a global leader in industrial AI, connecting operational data, human expertise, and AI to accelerate better decisions and measurable outcomes. By capturing and scaling institutional knowledge, Seeq helps industrial organizations improve reliability, efficiency, sustainability, workforce productivity, and profitability while keeping human judgment central to every critical decision.