This post explains why so many warehouse automation problems, and poor financial performance of automation projects, have little to do with the technology itself—and everything to do with poor process, bloated inventory, and lack of root-cause analysis.
A couple of years ago, I visited the warehouse of a regional logistics operator. A well-known automation company (better left unnamed) had invited a group of prospects—including several major automakers and national grocery chains—to a site visit showcasing this warehouse as one of their recent automation project success (spoiler alert: I think it was anything but a success). The agenda was exactly what you’d expect: vendor presentations, a tour of the automated warehouse, and lots of catered coffee and finger food.
At one point during his presentation, the warehouse manager proudly explained that the newly ramped-up facility we were about to tour was already operating at full capacity: the warehouse space was used up, and the miniload cranes were running at maximum throughput.
I made a mental note.
Because when someone says their system is running at “full capacity” just months after go-live, that’s rarely a good sign.
One Year Later
Roughly a year later, I returned to that same warehouse, but this time in a different role: as a professor accompanying a student group assigned to analyze the operation. What we found confirmed my initial suspicion.
The overall fill level of the warehouse was extremely high. The automated miniload system—designed as a goods-to-person solution with double-deep storage—was so full that it essentially deadlocked. The stacker cranes couldn’t retrieve totes from back positions because there was no room to shift front-position totes out of the way. Nor could they store inbound goods—there simply wasn’t space available. The system was paralyzed by overutilization.
Meanwhile, on the manual side, workers had to make picks from the top levels of the pallet racking—some of them more than ten meters high. This wasn’t just very inefficient. It was very uncomfortable to the employees. Several employees reported feeling uneasy leaning over from elevated platforms to reach into pallets so frequently. Curiously, the lower levels of the racks weren’t busy at all. Accordingly, productivity clearly wasn’t up to par.
We weren’t dealing with a hardware problem. The issue was poor storage allocation—products were being placed wherever space was available, not based on ABC logic or ergonomic considerations. In fact, any form of organized storage strategy had collapsed under the pressure of overstocking. Worse still, it wasn’t even a capacity problem. The warehouse was storing a massive number of SKUs with virtually no turnover. Based on some months of order data, the students found that only a small percentage of SKUs (low single digits) sold daily, and much (thousands of SKUs) had not sold in more than year. In other words: this wasn’t just a full warehouse—it was full of dead stock.
Band-Aid Automation
Here’s the uncomfortable truth: many warehouses are not well-managed.
- They hold inventory that has no business being there.
- Productivity isn’t tracked, or if it is, no action is taken.
- Turnover is high because workers are frustrated.
- Supervisors are untrained and unprepared to manage people.
Automation doesn’t solve these problems. At best, it hides them. It’s a band-aid solution—useful only if you’re treating minor cuts. But if you’re slicing your fingers open every day, applying a band-aid each time isn’t a fix. It’s a ritual. A distraction. A performance of problem-solving. A project theatre.
Spending millions on automation in a warehouse plagued by poor inventory control, absent data governance, and silly processes is impressive, but pointless.
This is not to say automation is unnecessary. But without clean processes, it’s pointless—and if your warehouse is already lean and well-run, it may be unnecessary.
The Universal Law of Warehouses: They All End Up Full
A former boss of mine once joked that logistics is governed by two universal truths:
- Every warehouse is different.
- Every warehouse will be full eventually.
He was right—especially about the second part. Humans have a deeply ingrained tendency to fill up whatever space is available. Basements, garages, suitcases, and, of course, warehouses.
The Lean community recognized this defect early on. Their response? Eliminate storage space wherever possible. Design out the temptation. Limit the buffer. It was one of their smarter moves.
In a different student project, we observed a warehouse worker trying to find a storage location for an inbound pallet. To be more precise: since no storage locations were available—which he already knew—he was scanning the racks for a pallet that had enough space so he could repack the arriving goods onto it. We stood there for about ten minutes watching him reconfigure pallets manually. This process was repeated many times per day, probably every day.
There’s no need to elaborate on productivity potential here—the losses were obvious. But the root cause was not labor inefficiency. It was the complete lack of storage capacity.
And if you’re trying to fix bad productivity with automation, without addressing the underlying reasons why your warehouse operates like this, then you’re just putting a band-aid on a wound you keep reopening.
You Can’t Fix a Symptom With a Robot: Why Root-Cause Analysis Comes First
Before investing in warehouse automation, the first question shouldn’t be “What can the system do?” but rather: What problem are we actually trying to solve?
Too often, companies jump straight to technology without ever performing a proper root-cause analysis. They notice symptoms—low productivity, long lead times, storage shortages, congested aisles—and assume that automation will fix them. But symptoms aren’t causes. And automating around a symptom may or may not make things better; if it does, the cost is extraordinarily high.
Let’s take high inventory levels as an example. It’s a common justification for automating: “We need to go vertical because we’re running out of space.” But high fill levels in a warehouse are rarely the real problem—they’re a manifestation of something else:
- Poor forecast accuracy
- Inefficient order quantities
- Procurement-driven volume discounts
- Excessive SKU proliferation
- Safety stock that was never recalibrated
- No coordination with sales or marketing on promotional volumes
- Slow-moving or obsolete stock that no one is managing
In the warehouse we studied, the storage system was completely overloaded—both the automated and manual zones. But the overload wasn’t due to strong sales or rapid growth. In fact, a high percentage of the inventory hadn’t moved in months. Some SKUs hadn’t sold in years. It wasn’t a storage problem—it was a decision-making problem in Procurement and a governance problem in Logistics. Had they automated based on observed fill levels, they would have automated their dysfunction.
This is why you need structured analysis before system design.
Sometimes, the problem is indeed inventory. But in other cases, the problem is capacity bottlenecks, slotting inefficiencies, unclear roles and responsibilities, poor inbound scheduling, or outdated master data.
We see this again and again in independent warehouse audits: teams are aware that “something isn’t working,” but they’re often looking in the wrong place. Fill levels are visible. Delays are visible. The underlying structural issues, however, require a bit more digging—and external perspective can be useful here.
And only after you’ve done that work—cleaned up your planning logic, reviewed your service strategy, restructured inventory policies—*then* you can ask whether automation makes sense.
Public Debate and Five Takeaways from the Field
This issue isn’t theoretical at all. A recent post I made on LinkedIn got quite a bit of reach:
A Norwegian business article from Finansavisen recently quoted Einar Øgrey Brandsdal, founder of the e-commerce giants Blivakker.no and Netthandelen.no. He argued that automation didn’t make financial sense for his company—and he’s right. His warehouses in Norway and Germany are already well-run, lean, and productive. Automating them would cost millions of euros and, at best, deliver marginal labor savings.
The headline of the article—“Successful Founder Tears AutoStore Apart”—was, of course, clickbait nonsense. Brandsdal wasn’t attacking AutoStore. He was making a valid business case argument. And his experience aligns closely with what I’ve seen in dozens of logistics projects across industries.
Here are some general points I’d like to add:
1) If you are operating a well-designed manual warehouse with solid software and well-thought-out processes, automation often won’t pay off.
2) Too many warehouse automation projects don’t make economic sense and actually increase the total cost per order.
3) In many cases, implementing a good WMS, redesigning some processes, and cleaning up your inventory will outperform a goods-to-person system—delivering better productivity and business results at a fraction of the cost.
4) Small-scale automation that targets specific pain points and a portion of SKUs often outperforms large-scale systems (such as goods-to-person picking) in both productivity and business case—again, at a fraction of the cost.
5) Productivity gains attributed to automation are often not primarily due to the automation itself, but to the fact that companies had to clean up their operations for automation to function at all—and usually acquired better software in the process. Most, if not all, of those gains could have been achieved without the automation.
And: Yes, some people have (rightfully) pointed out that labor shortages may justify automation even where there are no cost savings. To be honest, however, my impression is that (a) the issue of labor shortages is often greatly exaggerated—after all, millions of people are currently being subsidized by the government, including tens of thousands of recent immigrants; (b) several of our key industries are going down the drain, setting free quite a few people, thanks to the same governments that are incentivizing people not to work, and (c) increasing wages for warehouse staff could go a long way toward mitigating the problem.
The Real ROI Comes from Solving Real Problems
Let’s be clear: automation doesn’t make your problems worse. But it is an extraordinarily expensive way to hide them.
If your warehouse suffers from structural inefficiencies—poor storage logic, unbalanced processes, outdated data, missing communication loops—automation won’t make those problems go away. It will simply move them out of sight. The warehouse may look cleaner and more modern. Dashboards may be updated in real time. Robots may buzz around smoothly. But underneath, the same misalignments persist.
And because automated systems are less flexible, they often lock in whatever dysfunctions were already present.
As the saying goes: if you automate a bad process, you don’t get a good process—you get a bad automated process. Or, as Bill Gates put it: “The first rule of any technology used in a business is that automation applied to an efficient operation will magnify the efficiency. The second is that automation applied to an inefficient operation will magnify the inefficiency.”
Worse still, that bad process becomes harder to modify once automated. Every change now requires software reconfiguration, testing, integration checks, and (almost always) third-party support. What used to be a 30-minute fix on a whiteboard becomes a 6-month change request.
This is why real ROI doesn’t come from automation. It comes from solving real problems.
It comes from:
- Cleaning up how procurement decisions are made
- Clarifying roles and ownership of master data
- Re-aligning logistics, purchasing, and marketing (that’s a big one, by the way)
- Defining rules for SKU lifecycle management
- Rebalancing service level expectations across departments
And yes—occasionally, it comes from automating targeted processes that are well understood and stable. But that should be the last step in the process, not the first. If you skip the analysis, you risk cementing poor decisions into highly automated workflows—and spending millions to do so.
Final Thought: Start With Root Causes, Not with Robots
If there’s one takeaway from this article, it’s this:
Don’t start with automation. Start with a proper root-cause analysis.
Look closely at what’s going wrong in your operation—missed deadlines, low productivity, poor utilization, full warehouses—and ask why. Then keep asking why until you reach the underlying issue. More often than not, the root cause will lie outside the warehouse: in planning, in procurement, in data, or in the lack of cross-functional communication.
Once the root causes are understood, define your objectives.
Do you want to improve service levels? Free up space? Reduce labor dependency? Increase throughput? Shorten lead times? Lower your cost per order?
Only then—when the causes are known and the goals are clear—does it make sense to ask: Which solution fits? That solution might include automation. Or it might involve cleaning up your inventory, implementing a proper WMS, improving your planning processes, or redesigning storage logic.
But unless you’ve done the diagnostic work first, any technical system you buy will be a very expensive gamble.
Don’t start with technology.
Start with clarity.
Start with causes.
Start with objectives.
