We've watched companies waste millions on case packing and palletizing equipment. Here's what separates the failures from the wins.
Most food producers, at some point, look at the end of their production line and think: there has to be a better way. And they're right. Case packing and palletizing by hand is brutal, repetitive work. People get tired. They get injured. They leave. Finding more people is getting harder every year.
So you start looking at automation. You talk to vendors. You get quotes. You see impressive machines at trade shows. And then, if you're not careful, you spend a lot of money and end up not much better off than when you started.
We've seen it happen. One company in the UK spent close to a million pounds on case packing equipment. Impressive machines. But they hadn't bought a case erector to feed those case packers. And they hadn't bought palletizers to handle what came out the other end. So they still had people doing all the hard work — erecting boxes by hand, stacking them on pallets by hand — while expensive machines handled the one step in between. The labour savings they expected never materialized.
This isn't unusual. We've met several companies that paid millions for custom-built systems from local integrators, only to end up with equipment that performed worse than standard solutions already available on the market.
The pattern is almost always the same: someone focused on one piece of the puzzle without thinking about the whole line.
End-of-line automation isn't really about buying the right machine. It's a systems design problem. And the companies that get it right tend to think carefully about a handful of things before they ever talk to a vendor.
This sounds obvious, but it's where the complexity lives.
A food production line rarely produces just one product. In convenience food, you might be packing sandwiches in the morning, salads after lunch. Each product comes in different shapes, sizes, and packaging types. Each customer — each grocery chain — will have their own spec. Marketing decided on a new packaging format. Your production has to deal with it.
In the egg industry, the variety shows up differently. You're dealing with cartons of 4, 6, 8, 10, and 12. Each can have different label and carton configurations.
Then there are the cases themselves. Customer A wants display cases of a certain type. Customer B uses reusable plastic containers. They'll each give you the exact pallet configuration they require, and as a supplier, you comply.
Before you start looking at equipment, map out every product-case-pallet combination your lines need to handle. Not just today — think about what might be coming in six months. This map is the foundation of everything that follows.
It's also the first thing any vendor will ask you for. Document your configurations thoroughly — product dimensions, case types, pallet patterns, quantities — and share them early. A good equipment supplier can tell you the throughput you'd get on each configuration, which feeds directly into the next question.
Here's something people consistently underestimate: the number of hours your automation actually runs determines whether it pays for itself.
Say your production plan for next week looks like this: two hours packing SKU 1 for Customer A in the morning, two hours on SKU 2 for the same customer, three hours on SKU 3 for Customer B, then finally SKU 4 for Customer B.
If your automation can only handle SKUs 1 and 2, you're running it for four hours out of a full day. The rest of the time, you're back on your manual lanes. If your equipment only runs half the time, your return on investment is half as good.
What if Customer B suddenly orders SKU 1? Are you covered?
A useful exercise: take your production plans from the past few weeks and lay them over the system you're considering. How many hours would the equipment have run? Which production runs wouldn't have been covered? Are those gaps acceptable, or would they have caused problems?
Flexibility isn't a nice-to-have. It's what determines whether your investment makes sense. Before committing to a system, stress-test it against real production data, not just the plan for next week.
One more thing worth considering: talk to your customers. Sometimes they can accommodate small changes on their end — a different case type, a slightly adjusted pallet pattern — that make a big difference for your automation coverage. Once you're automated and more efficient, you might even be able to offer them a better price on the configurations your equipment handles well. That's a conversation worth having early.
And when you're thinking about runtime, remember that automation doesn't take breaks. It doesn't stop to accept a delivery or have a conversation. If your manual packers work an eight-hour shift, they're not packing for all eight hours. They take breaks, they rotate, they slow down toward the end. A machine running at a steady pace for the full shift will often outperform a faster manual process simply through consistency. When evaluating throughput, think about both the rate and the total runtime — it's the combination that determines your actual output.
We've gone through dozens of different layout options across our projects — different combinations of case packers, destackers, case erectors, printers, and palletizers. Each step in the line can be centralized (shared across multiple lanes) or placed inline (dedicated to a single lane).
The difference between the best and worst layout option can mean a lane needs to be attended to four or five times more often. That's not a rounding error. That's the difference between a smooth operation where someone checks in occasionally and a hectic one where someone is running around full-time.
The cost differences can be dramatic too. A centralized configuration — say, one case erector feeding multiple packing lanes instead of a dedicated erector per lane — means buying fewer machines. You get better utilization from each piece of equipment. The cheapest layout can cost half as much as the most expensive one, not because the individual machines are cheaper, but because you need fewer of them. Sometimes the integration is more complex with centralized setups — more conveyors, more routing logic — but the net savings are usually significant.
A few things to think about beyond the equipment itself: How much floor space do you actually have? If space is tight, you might be limited in what equipment can fit, which narrows your options.
Can your workers still move freely? If a conveyor blocks the path between two areas, someone now has to walk around a large machine every time they need to get to the upstream lane. That adds up. Passovers — walkways or bridges over conveyors — can solve this, but they need to be planned from the start.
Can forklifts access the pallets easily? Sometimes more space around the palletizer means the forklift driver can move faster and more safely. The layout around the palletizer often gets overlooked until it's too late.
When a person is packing by hand and a faulty product comes down the line, they just set it aside. An open egg carton? They close it. A badly printed label? They reject it. A leaking seal on a convenience food tray? Out it goes.
With automation, you need a plan for this. A rejection module where faulty products get pushed aside, or a robot that can set them apart.
One egg facility learned this the hard way. Open cartons — where the lid hasn't closed properly — are common in egg packing. It happens. Their case packer would detect the open carton and stop the line until someone walked over and removed it. This happened often enough that the lines were standing still for a significant portion of the day. They eventually took the case packers offline entirely and had them redesigned.
Before you buy, ask: what happens when a bad product shows up? How many can you buffer before the line stops? If rejection events are common in your process — and in food, they often are — this isn't a minor detail. It's the difference between a system that runs and one that doesn't.
Food traceability is driven by both regulation and retailer requirements, and the specifics vary by product category. Some operations track at the pallet level. Others go down to the box or even the individual product.
Automation can make traceability easier. For a human worker, traceability is one extra manual step — scanning a box, recording a lot number — which they might forget about under time pressure. A robot can scan every single box and report it to the traceability system automatically, every time, without exception.
Some retailers require human-readable labels on every box. If yes, you'll need a print-and-apply system, which adds equipment and cost. If your traceability requirements go down to box or product level, that's actually an argument in favour of automation — the consistency and speed can justify it.
Case packing gets a lot of attention, but palletizing is where a surprising number of projects go wrong — not because the palletizer fails, but because it wasn't thought about in relation to everything else.
We've seen companies spend three quarters of a million on case packers while still palletizing by hand. The palletizer, in many cases, would have been cheaper and delivered more immediate relief. It's the cost of omission rather than commission — people optimize for what's in front of them and forget about the broader picture.
When evaluating palletizers, the main questions are: How many pallet bays do you need? How many different SKUs does your production plan require you to palletize at the same time? What changeovers are required? What's the footprint? Are new pallets fed automatically, or does a forklift driver exchange them?
There are many different technologies — industrial robots, SCARA arms, Cartesian systems, robots on linear guides. For most operations, the fancier features like automatic pallet buffering and feeding are overkill. What matters more is whether the palletizer can keep up with your peak throughput and how it connects to the rest of your line.
And here's the throughput question again from a different angle: if your palletizer can't handle peak output, do you have a plan for overflow? A manual station? Buffer conveyors? Or do products just back up until someone notices?
Full automation in the sense of "nobody ever touches the line" is rare. Someone needs to feed in new cases. Someone needs to swap out full pallets. Someone needs to check on things.
The real question is: how often, and what happens while they're busy?
This comes down to buffering. If a palletizer gets stuck and there's no buffer, the line stops immediately. If there's a five-minute buffer, someone can finish what they're doing and walk over calmly. The first scenario creates a hectic work environment and dents your productivity. The second keeps things smooth.
Think about every point on the line where human intervention is needed and ask: how much buffer time do I have before the line stops? Then ask: is that enough given how many other things that person is responsible for?
The best automated lines we've seen aren't the ones where nobody shows up. They're the ones where someone checks in occasionally, feeds in material, and otherwise has time to accept deliveries, do maintenance, or take a break. Their day looks completely different from someone who spent eight hours packing cases by hand.
If there's one theme that runs through almost every mistake we've seen, it's this: spending a lot of money does not guarantee the right equipment.
Custom-built systems are expensive for a straightforward reason. The integrator has to cover the full cost of design, prototyping, and mistakes from a single customer sale. When things go wrong — and in a first-of-its-kind build, things always go wrong — you're paying for the learning curve.
An off-the-shelf machine spreads that development cost across many customers. The manufacturer has dozens or hundreds of units in the field. They stock spare parts. They have a support team that knows the machine inside out. When you need a part or a fix, it's routine, not a special project.
Custom systems have the opposite profile. Spare parts get made when needed, if the integrator is still around and if they don't have other priorities. Long-term support depends on the continued existence and goodwill of a company that might have a dozen other projects competing for attention.
This applies to software too. One facility spent €150,000 over several years developing a custom traceability system with a local software company. After all that time and money, they still didn't have a working system. The developer charged by the hour, not by the outcome. The company eventually gave up and bought an off-the-shelf system, Ovotrack, that worked and cost a fraction of the custom build. The same dynamic plays out with hardware.
Custom really only makes sense in niche industries where there are a handful of large players, each trying to build a proprietary edge. In food production, that's rarely the case. There are plenty of machine manufacturers building solid, well-supported equipment for exactly these applications.
And here's the other thing about off-the-shelf: when your product range changes — and it will — making adjustments is typically straightforward as the vendor supports it. With a custom system, every change is a new project.
A well-scoped project using off-the-shelf equipment typically pays for itself in one to three years. Multi-shift operations can often see sub one year returns on investment.
But these numbers only hold when the system is designed as a whole. The equipment cost is the obvious part. The less obvious part is everything else: the choices that determine how often someone needs to attend to the line, the flexibility that determines how many hours per day the system runs, the exception handling that determines whether it keeps running when something goes wrong.
Budget 5–15% of the equipment purchase price annually for maintenance and servicing. This is something people tend to think about late in the process, once they’ve already committed to a vendor. Better to factor it in from the start.
And remember to look at the full picture when building the business case. You might be able to build a system that handles every configuration. That's great, but if it costs three times as much as a simpler system that handles 90% of your volume, the math might not work out.
The best automated facilities we've visited have something in common beyond efficient production. The people who work there are proud of their operation. They're constantly looking for ways to improve. They feed in new cases every once in a while, watch the machines do the heavy lifting, and spend their time on work that actually adds value — maintaining equipment, managing deliveries, solving problems and talking to customers.
Nobody misses packing cases by hand.
End-of-line automation, done right, isn't about replacing people. It's about giving them better work to do while making your production more consistent, more traceable, and more resilient. The companies that get there are the ones that took the time to think it through before they bought anything.
Kenneth is the CEO of Witty Machines. He earned his PhD at ETH Zurich, where he taught robots to perceive and interact with objects using a few examples. Before that, he was an early engineer at Wolt and Webflow.
He's from Helsinki, Finland. He likes simple machines and analog technology.
