Warehouse Robot ROI: What Actually Earns Its Keep | Socius24
Warehouse Robots That Earn Their Keep vs the Ones That Don't
Over the last few years, a lot of money has been invested in warehouse automation that didn’t quite work out as expected. Significantly less has been spent on automation that did.
So, in this article, we’re going to talk about the difference between those two alternatives. We’re not going to talk about whether or not to automate warehouses, because anyone with any kind of competitive edge already knows the answer to that one: if you want to stay relevant, you’ve got to do something.
The more important question, though, is which kind of warehouse automation offers good Return on Investment (ROI), and what its implementation could do for the people currently doing the jobs that it’s aimed at.
The $290 million bet on warehouse robots
Way back in 2021, at the height of the trend for autonomous mobile robots (AMRs), one of the better-known names in enterprise hardware paid around $290 million to acquire an established AMR maker. These were the kinds of robots that trundled through a warehouse alongside pickers, cutting down on walking and speeding up picking. The buyer evolved their AMR purchase into a branded fulfilment offering, kept on investing and added new features.
However, in December last year, that same company filed to wind down its whole robotics unit, reporting that it was, instead, re-sharpening its focus on core business. The market intelligence people had a somewhat blunter interpretation of this, which was that the AMR market simply never turned into the silver bullet that everyone buying into it in 2021 had assumed it would become.
I am not pointing this out to be unkind. The company in question makes excellent kit, and it clearly knows its core business well enough to know when something isn’t it. I am pointing it out because this is the most reliable signal that the industry has given us in a while: a serious, experienced, international operator looked at its robotics, ran the numbers, and then decided that the returns just weren’t worth their time and investment.
Which rather contradicts the prevailing vibe that any warehouse not currently knee-deep in robots is somehow falling behind.
The word "AI" is doing a lot of heavy lifting right now
Rather than being one, single, tangible thing, “AI” has become a somewhat generic term that people stick somewhere on a presentation slide when they want their budget request to go through without too many questions.
But it’s a spectrum.
There’s genuinely clever, pinpointed software at one end of that spectrum, and then there are glorified rules engines that have spent their budget on good marketing (rather than effective development) at the other. And worryingly often, buyers are left to work out which one of those things they’ve bought. After it’s been installed.
Most of the chatter in the market right now is coming from the vaguer end of that spectrum. And any platform that promises to use AI to transform your entire operation is usually promising a roadmap, rather than a specific product.
But where the technology is truly delivering right now is at the opposite, rather more specific end: AI is earning its keep most effectively when directed at one, well-defined, repetitive problem that humans find tedious and computers find easy. And then, it’s being left to get on with it.
Where warehouse automation IS working
GNC, the health and supplements retailer, had an inventory counting issue. Counting 40,000 locations across 450,000 square feet meant taking their lift trucks and staff off the floor, working weekends, and still only managing a couple of full counts a year. Which, in a building that never stops, is a proper headache to have to deal with.
So, instead, they put autonomous drones in to do their counting. The drones fly at walking pace, read the barcodes, and feed the counts back. Full counts went from two to four times a year up to ten to twelve. Accuracy improved and discrepancies that used to take months to come to light started to get caught much more quickly. The inventory control team went from twenty people to thirteen, and the other seven people were moved into work that needed the kind of exception-handling and problem-solving that a drone just cannot provide. GNC’s distribution chief described their decision rather nicely: they chose to back technology that actually did the job, rather than choosing technology that looked impressive while it was doing it.
Get your skates on...
The second example of this, I like even more, because of where it came from. At IKEA’s Helsingborg site in Sweden, a member of the floor staff saw an advert for a pair of robotic boots called Moonwalkers. The boots fit over ordinary footwear, using adaptive software to match the wearer’s stride. Staff ordered a pair and tried them out. And when they did, their pickers ended up walking up to twice as fast and picking productivity per hour rose by about sixteen percent. IKEA’s own framing of this was that they were investing in tools for their workers that make it easier to work smarter rather than harder.
Which is about as good a one-line summary of the whole argument as I’ve seen.
Please note what both of these technologies have in common: nobody bought a grand vision. Instead, someone identified a specific, irritating, repetitive task and found a tool that did THAT.
They didn’t get rid of their people. They just found tools that made their days easier or more useful.
Garbage in, more confident garbage out
There’s a catch that AI optimisation pitches tend to skip over: none of this clever software does you any good if the data underneath it is wrong. The Auburn University RFID Lab puts average inventory accuracy across US retailers at around sixty-five percent, which means more than a third of the data many operations are running on is just… well, wrong. And if you point your shiny optimisation engine at numbers like that, all it’s likely to do is optimise its way (with the kind of confidence of a 4-year-old wearing a superman t-shirt) right into the nearest wall.
And this is a very overlooked part of optimisation.
Which is ironic, because it’s the part that we care the most about. A warehouse management system that’s well implemented and properly run is the foundation that everything else can confidently be built on top of. Blue Yonder Dispatcher WMS users typically report inventory accuracy that is close to one hundred percent. And fundamentally, that statistic is the difference between optimisation tools that work and optimisation tools that just make things worse, more quickly and effectively.
Where we're seeing warehouse optimisation ACTUALLY pay off
As soon as your WMS foundation is solid, narrow, well-aimed optimisation tools really start to come into their own.
The example that’s closest to home for us is Optioryx Pulse, which we offer through our technology partnership with Optioryx. It optimises picking, and does it very, very well. It works out smarter sequences and shorter routes based on what’s actually happening on your floor (rather than, say, rules you might have set up a while back and not updated), so that the same picker can cover less ground and get more done.
Pure-automation gambles can come unstuck. And vague AI promises often underdeliver. But what works, actually, reliably, currently, is a solid WMS that underpins everything, accurate data, and focused tools that are aimed at the kinds of jobs that are slow, repetitive and expensive.
That combo results in freeing up people for the kind of work that needs a person, rather than replacing people with a machine that half does their job. And it’s also, not incidentally, how you get to keep good staff… rather than losing them to your competitor down the road who treated automation as a way to make their work better, rather than a way to make their payroll smaller.
If you’re weighing up where automation or optimisation might help your operation, compared to where it might instead be buzzword bingo, we’re always happy to talk things through. Book an obligation-free discovery call now.
FAQ: Warehouse automation ROI
Narrow, focused tools aimed at one specific, repetitive, expensive job. Inventory-counting drones and picking optimisation software are consistently outperforming grand “transform your whole operation” platforms, because they solve a defined problem and can prove it with numbers.
Accurate data, which in practice means a properly implemented warehouse management system. Average inventory accuracy across US retailers sits at around sixty-five percent, and optimisation software pointed at bad data simply makes mistakes faster. Blue Yonder Dispatcher WMS users typically report inventory accuracy close to one hundred percent, which is the foundation these tools need underneath them.
A picking optimisation tool that Socius24 offers through its technology partnership with Optioryx. It sits on top of Blue Yonder Dispatcher WMS and works out smarter pick sequences and shorter routes based on live floor activity, so pickers cover less ground and get more done.
Request a Demo Now
If you’ve enjoyed this article, claim your free subscription to our LinkedIn weekly Newsletter
– The World of WMS –
for more of the same.
