Automation is becoming one of the main investment areas in packaging and manufacturing. Esko’s 2026 Packaging Trends Survey placed automation ahead of AI among the technologies expected to have the biggest impact on packaging. PMMI’s 2026 Robotics in Packaging and Processing report also found that 72% of surveyed end users already use robotics in packaging and processing operations. 

So the direction is quite clear. But from an engineering point of view, the real question is not whether to automate. It is where automation actually creates operational value.

In packaging lines, this distinction is important because line performance is not simply the sum of the rated speeds of individual machines.

Let’s take a simple example. A filler runs at 200 units per minute. The next process can handle 170, and the end-of-line system effectively runs at 140.

In this case, installing a faster filler does not automatically increase the output of the line. The real limitation may be somewhere else: downstream blocking, upstream starvation, insufficient accumulation, frequent micro-stops, changeovers or manual handling between processes.

This is why I prefer to start with system throughput rather than machine speed.

Machine OEE is not line performance

OEE is still a very useful metric, but it mainly gives us an equipment-level view. A packaging line is a connected system. Because of this, the machine with the lowest OEE is not always the machine that limits total line output.

A study on a beer filling line showed exactly this point, the equipment with the lowest efficiency was not necessarily the unit limiting total line productivity when overall throughput effectiveness was considered. 

This is easy to miss in real production environments. Several machines may all have acceptable OEE values, while total line throughput is still poor because of speed mismatch, buffering problems or interaction between processes. For me, the objective should not be to make every machine run at maximum speed.

The objective should be to get the highest stable output of good product from the complete line.

Packaging material is also a process variable

Another point that is often overlooked in automation projects is the packaging material itself. A film, carton, bottle or closure is not only a purchased material. Once it enters the production line, it becomes a technical process input.

Changes in coefficient of friction, stiffness, dimensional tolerance, sealing range or hot-tack behavior can affect feeding, tracking, sealing and achievable machine speed. Research on form-fill-seal packaging has shown the relationship between packaging-film properties, heat-seal behavior and packaging performance.

This means a material change may reduce purchasing cost but still increase total conversion cost if it causes more stops, rejects or speed losses. So, when a line problem starts after a packaging change, checking machine settings is not enough.

We also need to ask whether the new material still gives the machine a suitable process window.

Hidden capacity should come before new capacity

Before a major automation investment, one of the first things I would check is how much capacity already exists in the line but is currently being lost.

Changeover is a good example. A food-factory case study using SMED and related lean methods reduced changeover time by nearly 30%, increased OEE to more than 70% and reduced labor cost by 10%.

Of course, this does not mean every factory can achieve the same result. But it shows something important: usable capacity can sometimes be released without increasing the technical speed of the main equipment. The same applies to micro-stops, poor line balance, unnecessary manual transfer, badly positioned inspection points and inadequate accumulation.

Before buying new capacity, it makes sense to understand how much existing capacity is hidden inside the process.

Automation does not always need to start with a robot

Automation is sometimes discussed as if there are only two choices: manual operation or full automation. In reality, there is a large engineering space between them. A sensor, mechanical guide, stopper, gravity transfer, simple diverter, interlock or small handling device can remove repetitive human intervention with much less complexity and investment.

I use the term “low-CAPEX automation” for this approach.

It does not mean zero investment. It simply means that before going directly to major equipment, we first look at process design, mechanical simplification and small control solutions.

This matters because the highest level of automation is not always the best solution for the process. PMMI’s 2026 From Complexity to Capability report specifically asks how manufacturers can scale capacity without adding unnecessary automation complexity. It also highlights the industry's shift from peak line speed toward mix-adjusted throughput.

Product mix, changeover frequency, maintenance skills, spare-parts strategy, flexibility and future volume all need to be part of the automation decision.

Deloitte’s 2025 Smart Manufacturing Survey also shows how important automation has become: 46% of respondents ranked process automation as a first- or second-priority investment for the next two years, while 37% did the same for physical automation.

The investment case is strong. But applying it in the right place is just as important.

Start with the constraint

Before a major packaging-line automation project, I believe the sequence should be simple.

First, find the real bottleneck.

Then separate availability, speed, quality, changeover, starvation and blocking losses. Check whether packaging material behavior is part of the problem. Review line balance and manual handling. Look for low-CAPEX improvements.

Only after that should the automation investment be defined.

The goal is not to automate as much as possible.

The goal is to automate the right part of the process, at the right level, where it creates measurable value in throughput, cost, quality or consistency.

When we start from that point, technology becomes what it should be: an engineering solution, not the objective itself.