Factories Keep Adding Technology. 58% of Workers Say They Are Getting Less Productive

For most of the past decade, the answer to a productivity problem on the plant floor has been a purchase order. A new scheduling module, a quality app, a dashboard, and now an AI pilot. A survey published this month asks the people who live with those systems whether the plant is actually getting faster, and most of them say it is not. The problem they describe is not the technology itself. It is that the work around it never changed.

The Tool Arrived. The Process Stayed the Same

The survey, commissioned by the consultancy Argon & Co and published on October 8, polled 500 workers and managers in Australia's food and beverage sector online through Zoho. 58.4% said their organisation is becoming less productive despite continued investment in workplace technology. More than three-quarters, 76.4%, said new technology is typically introduced without changing the processes, responsibilities or broader ways of working that surround it.

Asked what stops them getting more from technology, about six in ten pointed to outdated processes or poor integration with existing systems, and 34% said the underlying processes are never updated at all. Only 14% blamed a lack of leadership support. "Technology alone won't fix an inefficient process," said Paul Eastwood, managing partner at Argon & Co. The respondents are not asking for less technology. They are describing systems that were switched on without anyone redesigning the job they were meant to improve.

The cost shows up as duplication. 95.8% said work is duplicated across systems, teams or processes, and 75.8% put that duplication at a moderate or significant level. Respondents described entering the same data into multiple systems, keeping spreadsheets running alongside enterprise platforms and manually reconciling production information. 81.8% spend more than a quarter of their working week on repetitive or manual tasks, and 40.8% said those tasks take up more than half of it.

AI Is Being Layered Onto the Same Old Work

AI has not yet changed that picture for most of these plants. 82.4% said their organisation has introduced AI only to a limited extent or not at all. Among the rest, results are mixed: 41.8% of respondents said AI brought no noticeable productivity improvement or made their work harder, while 14.4% reported a significant boost.

The broader industry data points to the same gap between promise and practice. Deloitte's AI in Manufacturing 2026 study, a survey of more than 140 manufacturers, found that 84% report measurable value from AI in their operations. Yet only around one in five use cases has been scaled consistently across sites or enterprise-wide. Among the implementation challenges respondents named, resistance to change and a lack of technical expertise were each cited by 35%, and data availability or quality by 30%.

Read together, the two surveys suggest the value of AI is real in the pilot cell and leaks away on the way to the rest of the plant. A model that sits on top of hand-reconciled records and a spreadsheet kept in parallel inherits every one of their errors, and asks operators to check one more screen on top of the ones they already have.

The Data Is Collected. Most of It Is Not Put to Work

The Argon & Co sample is Australian and confined to food and beverage, but the pattern is not. Rockwell Automation's 2026 State of Smart Manufacturing report, based on 1,560 respondents across 17 countries, found that 59% of manufacturers actively use smart manufacturing technologies and only 18% remain in pilot mode. One-third of operating budgets are still dedicated to industrial technology. Yet manufacturers say only 43% of the data they collect is used effectively. Rockwell's own conclusion is that execution, not data availability, is now the constraint.

The spending is not slowing down. In the Philadelphia Fed's October Manufacturing Business Outlook Survey, 39.1% of firms expect to spend more on software in 2027 than in 2026, against 8.7% expecting to spend less, and 52.2% expect total capital spending to rise, with none expecting a fall. That is another round of new systems heading for the same plants. Unless the processes change with them, it is also another round of duplicate entry, parallel spreadsheets and manual reconciliation.

What Operations Leaders Should Do Now

• Map the work before buying the tool: follow one batch or order from schedule to shipment and count every point where the same data is keyed, checked or reconciled more than once. That count is the baseline any new system has to beat.
• Retire something with every launch: each new system should switch off a named spreadsheet, paper form or manual check, written into the project plan and confirmed after go-live, not left running "just in case".
• Give every process an owner: system owners keep software running, but someone has to own how the work flows through it and have the authority to change roles and handoffs when the tool arrives.
• Measure the floor's time, not the licence count: track the hours operators and supervisors spend on repetitive and manual tasks before and after each rollout, and treat a flat number as a failed deployment.
• Point AI at a named bottleneck: 63.2% of the Argon & Co respondents saw the biggest opportunity in production, supply chain and inventory management, and 52.2% put reducing repetitive tasks, errors and rework first. Start there, with a measurable problem.
• Use the data you already have: before adding sensors or platforms, find the reports nobody reads and the data nobody acts on, and either connect them to a daily decision or stop collecting them.

The plants that get faster from their next technology budget will not necessarily be the ones that spend the most. They will be the ones that redesign the job before the system goes live, and switch off the old way of working the day the new one starts.