A factory owner tells me, with total conviction, that our technology is exactly what his business needs. And then: nothing happens.
Not resistance. Not objection. Enthusiastic agreement followed by silence. “I’ll schedule the onboarding meeting this week.” A week passes. “Sorry, things have been busy.” Another week. The dashboard sits ready, waiting for data that never arrives.
This is not one factory. It is three. And what this pattern reveals about how change actually works applies far beyond garment manufacturing.
I run DAITA, bringing AI-driven operations management to factories in Tiruppur, southern India. Our clients are not technophobes. They have spent hours in research sessions with us. They see clearly that their current systems are broken: merchandisers manually copying data between spreadsheets, firefighting problems that could have been prevented. They want the fix. They just never quite start it.
The resistance is not vocal. It lives in deprioritisation: the meeting that gets rescheduled, the “I’ll call you in an hour” that becomes five hours of silence, the worker who says “my orders haven’t come in yet” without realising the system is designed to start before the crisis, not during it. Our system asks people to be proactive, to do work now that prevents problems later. It turns out that asking people to act before there is a fire is one of the hardest behavioural changes you can make.
I watched one factory delay because they had too many orders: “Not the time to change anything.” Another delayed because they had too few: “Nothing to track yet.” The excuse shifts with conditions; the outcome does not. That tells you the barrier is not circumstantial. It is structural.
What they are protecting
When you bring AI into a factory, there is a reality that rarely gets spoken aloud: this technology can do a portion of someone’s job. A 40-year-old merchandiser who has been doing this work for fifteen years does not care about your macro argument that AI makes people more valuable. From where he sits, the calculus is personal. His routine works. His skills are proven. He can move to another company tomorrow. The owner bears the risk of the company failing; the merchandiser bears the risk of looking incompetent in front of his colleagues while learning a new tool.
That is not stupidity. It is a perfectly rational response to a set of incentives. Kahneman and Tversky’s research on loss aversion formalises what any factory floor will teach you in a week: the pain of giving up what you have feels roughly twice as heavy as the pleasure of gaining something equivalent. The familiar, even when it is worse, feels safe. The new, even when it is better, requires stepping into uncertainty where competence is temporarily in question. And the power dynamics compound this. Workers on the factory floor have more informal power than any org chart suggests. Owners are reluctant to push too hard. Conviction at the top does not automatically become action on the floor.
What we changed
We stopped waiting for factories to onboard themselves.
First, we stopped selling the upside. “DAITA will help you manage your factory better” moves nobody. “Every month you track production manually, missed deliveries, idle time, and rework you cannot see are costing you hundreds of thousands of dollars a year.” That moves people. Within two weeks, we give the owner something he has never had: his entire team’s performance on order delays, both supplier and buyer, visible in one dashboard, accessible from his phone, with no manual data entry. When the technology makes the hidden cost of not changing impossible to ignore, it becomes its own argument.
Second, we stopped assuming enthusiasm would cascade into follow-through. We now name two roles from day one. A project sponsor, typically the factory owner, who holds both teams to a calendar. And a project owner on the client side, responsible for driving daily adoption. Below them, a handful of change champions: the people who will actually use the system. We meet them daily until the workflows stick. Not weekly. Daily. Because the gap between “I have been trained” and “I do this automatically” only closes through repetition, not instruction.
Without that structure, we were trying to manage someone else’s internal change from the outside. It does not work. The person who pushes adoption has to sit inside the organisation, not visit it.
The ones who adapt are the ones who win
ThreadSol, a Delhi-based startup now part of Coats Digital, cracked a similar problem in Indian garment factories a decade ago. They targeted fabric waste in cutting rooms, where material represents 70 to 80 percent of manufacturing cost. What worked was proving ROI in rupees, not features, and making the interface simple enough for any worker to use. The factories that adopted early gained a measurable edge. The ones that waited kept bleeding margin without knowing where it went.
That competitive pressure is building again. Factories that adopt data-driven operations can spot delays before they cascade and give international buyers the transparency they increasingly demand. As competitors in Bangladesh and Vietnam modernise, “we have always done it this way” stops being a comfort and starts being a liability.
There is a version of this story that is just about a startup learning how to onboard customers. But the principle underneath is universal. People do not adopt tools because they are better. They adopt tools when the pain of not adopting finally exceeds the effort of changing. Your job is to make the pain visible and the effort invisible.
Enthusiasm without accountability is just goodwill.
