How to get your shop floor to actually use new software

This guide is in three parts. Jump straight to the one you came for.

01Why do software rollouts fail in fab shops?

They fail on people and process, not technology. The numbers people quote at you come from big ERP rollouts, and those miss what they set out to do somewhere between 60 and 75% of the time, depending who is counting. The story behind the number is always the same: the system works on paper while the crew quietly works around it. Those horror stories are real. They are also a different animal to what we are talking about here - a system the size of a whole business, dropped on a shop that never asked for it.1

Here's the part that should give you confidence rather than dread. A firm called Prosci has been keeping score on this for 25 years. When somebody actually runs the change, the job is about seven times more likely to land - close to nine out of ten of them hit or beat what they set out to do. And pushback isn't some freak event. At the big outfits that measure this sort of thing, the people running the project name staff pushback as the biggest thing in their way more often than anything else, and over half of them run into it. Different world, same lesson: if your crew grumbles, you are not the exception. You are the rule.2

Two things separate the shops where software sticks from the shops where it dies. The tool has to be built for the work - a bench, gloves, noise, no patience for menus. And the change has to be run, not announced. Factory is deliberately the smaller, simpler system: we set it up in stages, we teach each person their own part, and getting your team using it is part of what we do rather than something left in your lap. Those numbers up there are what we built against.

02Whose job is it to make people use it?

Almost all of it is ours. One part is only yours: backing it out loud, and not going quiet on it later. The research has had that at number one for 25 years running - ahead of budget, ahead of the software itself. It's also the only part you can't hand to someone else. The rest of this list is our job.

If you read that and thought "the left column is five conversations, not five weeks of work" - that's the point. Your job is the decision and the backing. Ours is the rest.

03The 20-year veteran

“My guys have done it the same way for 20 years. They won't touch software.”

Old hands aren't anti-technology. They're anti-being-made-to-look-slow in front of people they've trained. Take the audience away and most of the resistance goes with it. That is why the training isn't one fixed format: we run group sessions where they work, and take someone aside one-on-one where they don't.

The floor is genuinely older than it used to be, and that's worth planning for rather than worrying about. The average manufacturing worker in the US is now around 44, a few years older than the workforce generally, and about one in four is 55 or over. Shops with a lot of older hands used to be unusual and now they are normal: the share of manufacturing jobs in workplaces where a quarter of the crew is 55-plus went from 14% in 2000 to over 40% by 2022. Here's the bit that matters. Someone went and asked older workers about training, and barely half had been offered any in five years - while plenty of them said they wanted it. Nobody was asking them.3

There's a second move that works better than any training plan: ask them to sanity-check the setup. Your senior fabricator knows which stages a job really goes through and what the statuses should be called. Bringing that knowledge into the build turns the person most likely to resist into the person whose fingerprints are on it. It costs you one conversation.

04Language on the floor

“Half my floor doesn't speak English as their first language. What happens to them?”

You plan for it deliberately instead of hoping it works out. The daily habit - clock on, pick up a job, mark it done - is taps and status colors, not reading. That's how it's built, not a workaround: the floor screens run on colors and pictures, on a phone or a tablet, because a bench is not a desk.

This is a normal shop reality, not an edge case. About one in ten American adults doesn't speak English well, and roughly the same share of the workforce. Manufacturing has more of them than most industries. Now look at what the employers themselves say: they list English as one of their biggest challenges, and in the same breath rate those same workers as productive and committed. Worth reading twice. The people were never the problem. The instructions were.4

• →Training happens on the floor, on real jobs. Muscle memory doesn't need a translator. Someone doing their own job on their own tablet, three times, learns more than an hour of explanation.
• →We pair a bilingual workmate into the session. Not as an interpreter for a lecture - as the person standing next to them for the first few jobs, and the person they ask afterwards.
• →The screen carries the meaning. A job's state is a color and a photo, and the action is a button. If a person can read the board across the shop, they can read the tablet.

Your job here is one sentence at the announcement: nobody is being tested on their English. The training that follows is on us.

05The quiet sandbagger

“My foreman will nod along in the meeting and quietly kill it on the floor.”

First, tell them before you tell anyone else. Most "sandbagging" is a good operator who found out last. But when someone has decided to wait it out, you need to be able to tell the difference between pushback and sandbagging - because they need opposite responses.

The signs are consistent enough to be worth naming: the spreadsheet reappears "just for now", there's never time for the thing that takes four taps, job updates only ever get entered by someone else, and the old way gets quietly recommended to newer staff.

One more thing that does most of the work here: pick the date the old way stops, say it early, and don't run two systems past it. An indefinite parallel system is an open invitation to wait it out. More on the leadership side of this in how to lead change without losing the floor.

06The bookkeeper

“So it writes the invoices into the accounts by itself. What is left for me to do?”

This one rarely gets said out loud, which is exactly why it gets missed. The person who does your books has spent years being the one who knows where everything is. Then you announce software that creates the records in your accounting package on its own. From where they are sitting, that is not a time-saver. It is a question about their job, and they will not ask it in front of the room.

So answer it before it gets asked. The typing goes. The judgement stays. Factory does not replace your accounting package, it feeds it, so the same job stops being entered twice. What that frees up is the work they never get to: chasing money owed, checking what jobs actually cost against what you quoted, spotting the work that is losing money while it is still on the floor, closing off the month without a late night. That is a better job than re-keying invoices, and most bookkeepers know it before you finish the sentence.

One honest note. If the plan really is the same team doing better work, say so plainly and early. If you cannot say that honestly, do not say it. They will find out either way, and everything else you told them stops counting.

Then give them the part that is genuinely theirs. Your chart of accounts, your tax codes, how jobs should map across: that is their knowledge, not ours and not yours. When the accounting connection is their piece of the setup rather than something done to them, the whole conversation changes. More on why one set of numbers matters in why your workshop needs a single source of truth.

07“I haven't got time”

“I haven't got time to run a software project on top of running the shop.”

You're not running it. We are, and we've done it hundreds of times. The same person owns your setup from kick-off to live, they know fabrication, and the heavy lifting sits with them: your customers, suppliers, products and pricing migrated, your kits and BOMs built out, your team trained role by role. Not handed to your already-busy crew as homework.

What you actually spend time on is the kick-off conversation about how your shop runs, a handful of decisions only you can make, and the announcement. That's it. Hundreds of shops have been through this, which means we're not learning on you. Whatever your version of "this won't work here" is, we've almost certainly seen it.

08Why the last one died

“We tried something before and it died. Everyone here remembers it.”

Then your team isn't the problem, and it's worth saying that to them out loud. Your team doesn't hate software. They hate software built for someone else's job. Around eight in ten people in the world work on their feet rather than at a desk. For decades, roughly 1% of the money going into new software went anywhere near them. So what turned up on shop floors was office software handed to people who don't sit down.5

What goes wrong is boringly specific. About half the supervisors surveyed said the number one hold-up is simply getting into the thing: logins, passwords, the wrong device. People on the floor want it on a phone, and short lessons of three to five minutes get finished two to three times more often than long ones. Meanwhile seven in ten of them say tech already makes their job easier. The willingness is there. It turns up when the tool fits the work.6

So the honest diagnosis of the thing that died here last time is usually the wrong tool and a change nobody ran - not a verdict on your people. That's worth naming at the announcement, because right now half your floor thinks the last failure was their fault, and the other half thinks it proves nothing new ever works here. See also: how to introduce tech to a team that's not tech savvy.

09Who's on your floor?

Six people turn up in nearly every shop we set up. Each needs a different first ten minutes - and in most cases the move that works is the opposite of the instinctive one. Click a card to see what works, what backfires, and who handles it.

Notice how few of these are about software. They're about being told first, being trained privately, and not being made to look slow. That's the whole game.

10What actually works, according to shops that have done it

This isn't our theory. The same three patterns show up wherever a fabrication floor successfully changed how it works - pair people up, teach each role only its own part, and treat it as a leadership job rather than an IT one.

• →Pair the comfortable with the experienced. The Fabricator ran a story on a mechanical outfit, Shapiro & Duncan, going paperless. They put the guys who were comfortable with a screen alongside the old hands - and it went both ways, with trade knowledge going back to the younger ones. Everyone was working in the new system early on, and the result showed up on the floor: job turnaround went from weeks down toward five days.
• →Involvement decides the outcome, not the technology. The Fabricator also wrote about Skilcraft putting the same automation in twice. The crew who got left out of the decision dug their heels in. The crew who were involved, and got weeks of hands-on training, turned that line into the job people wanted.
• →Start with a small crew, and walk each role through their part only. The advice that gets repeated across the fabrication trade press is blunt: start with a few people, show the welder the welder's screen and the loader the loader's, and treat it as a leadership job rather than an IT job.

Read that list again and you're reading how we get a shop set up. In stages, each person taught their own part, hands-on, paired up, and signed off on real work instead of a training sheet. We built it that way because it's the only version that survives a fab shop.