There’s a particular kind of silence that falls over a production meeting when someone asks a simple question and nobody in the room can answer it.

You know the one. Somebody from sales wants to know why the order that was promised for Thursday is now going out the following Wednesday. And the room goes quiet. Not because people are being difficult, but because the answer lives in about four different heads, none of which have spoken to each other since last Tuesday. Maintenance knew the machine was struggling. The night shift supervisor mentioned it to somebody. Quality flagged a dimensional drift a fortnight ago and it went into an email that got buried. Planning never heard any of it.

Nobody did anything wrong, exactly. And that’s the problem.

Here’s the thing about a disconnected team. It never appears on your profit and loss statement. There’s no line item that reads “cost of people not talking to each other.” You’ll find energy costs, materials, labour, depreciation, insurance. You won’t find the £300,000 that quietly walked out of the building last year because your shift handovers are a scrap of paper and a shrug.

That invisibility is exactly why it survives. Anything you can’t measure, you can’t defend a budget against. So when you go to your board or your MD and ask for money to develop your supervisors, or to fix how your teams work together, you’re competing against a new CNC machine with a quoted payback of eighteen months and a supplier brochure full of confident numbers. You’ve got a feeling. They’ve got a spreadsheet.

To be honest, I think that’s the real reason so much of this gets kicked down the road. Not because manufacturing leaders don’t believe in it. Most of the ones I’ve met believe in it deeply. It’s because they’ve never been handed the arithmetic.

So let’s build the arithmetic. By the end of this you’ll have a number. Your number, from your plant, using your data, that you can put in front of a finance director and defend.

Disconnection doesn’t announce itself

Before you can cost it, you need to recognise it. And it rarely looks like conflict.

I spent a morning a while back at a precision engineering firm in the West Midlands. Roughly 140 people, decent order book, third generation family ownership, the sort of business that’s genuinely good at what it makes. Walking the floor, everything looked healthy. Clean, organised, people busy and reasonably cheerful.

Then I sat in on the eight o’clock meeting.

The production manager ran through the day. Maintenance sat at the far end of the table and said almost nothing. When I asked afterwards why that was, the maintenance lead gave me an answer I’ve thought about ever since. He said he’d stopped raising things in that meeting about two years ago, because whenever he did, it turned into a conversation about why he hadn’t fixed it yet. So now he just gets on with it, and tells production when it’s done.

That’s disconnection. Not shouting. Not sabotage. A capable, conscientious person quietly deciding that sharing information isn’t worth the aggravation.

Once you start looking for it, you see the pattern everywhere:

Information that only travels through one person, so when they’re on holiday the whole thing wobbles. The WhatsApp group that’s become the real production system, running parallel to whatever software you’ve paid for. Sales agreeing to lead times without asking operations, and operations assuming sales will always overpromise, so they pad every estimate as a defence. Quality being treated like the police rather than a resource. Two shifts running the same cell with genuinely different methods, and neither of them knowing the other exists.

And the classic. Somebody spots a problem early, mentions it once, gets no response, and never mentions it again. That one costs more than all the others combined, and you’ll never know it happened.

Gallup has been running the largest ongoing study of workplace performance for decades now, covering well over 180,000 business units. One of their most consistent findings is that roughly 70% of the variance in team engagement is explained by the quality of the manager. Not pay. Not perks. The person your people report to. Which means disconnection is usually not a workforce problem at all. It’s a capability gap sitting in your supervisor layer, and supervisors in manufacturing are very often brilliant technicians who got promoted and never received a day of training in how to lead.

That’s not a criticism of them. It’s a description of how our industry has recruited for about fifty years.

The four places the money actually leaks

Right. Numbers.

I’d suggest four buckets, because four is enough to be credible and few enough that you’ll actually complete the exercise. Missed deadlines. Preventable mistakes. Morale and absence. Turnover and lost knowledge.

Missed deadlines and disruption. Not all downtime is a coordination failure. A bearing fails, a bearing fails. But a surprising slice of it is. The changeover that took ninety minutes instead of thirty because the tooling wasn’t staged. The line that ran the wrong drawing revision for half a shift. The stoppage that maintenance could have prevented if anybody had told them about the noise.

Research from Censuswide for Fluke in 2025 found that around 68% of UK manufacturers had experienced unplanned downtime in the previous twelve months, with nearly half reporting six to ten incidents a week. Separate work from IDS-INData suggests a typical UK company loses something in the region of 49 hours of production a year to it. I’d treat the eye watering per hour figures that circulate with these studies carefully, since they’re drawn largely from very large plants and commissioned by suppliers who sell the fix. But the frequency data holds up. Six to ten stoppages a week is a lot of small fires, and a decent proportion of small fires are lit by somebody not knowing something somebody else knew.

Preventable mistakes. The American Society for Quality has long put the cost of poor quality at somewhere between 15% and 20% of sales revenue for many manufacturers. Scrap and rework on their own run from around 0.6% of revenue for strong performers to about 2.2% for weaker ones. And the honest analyses tend to land at three to five times the visible scrap figure once you count the machine time consumed, the capacity lost, the engineering hours spent investigating and the expedited freight to cover the gap.

Look at the root cause lists for scrap and rework and you’ll find the same culprits every time. Human error. Unclear or outdated work instructions. Inconsistent processes between shifts. None of those are equipment problems. All of them are communication problems wearing an engineering costume.

Morale and absence. The CIPD’s 2025 Health and Wellbeing at Work report found UK employees taking an average of 9.4 sick days a year. That’s up from 7.8 in 2023 and 5.8 before the pandemic, and it’s the highest they’ve recorded in over fifteen years. Mental ill health was the leading cause of both short- and long-term absence, and 64% of organisations reported stress related absence with workload as the top factor.

Here’s the part that ought to sting. Only 29% of organisations train their line managers to support staff with mental ill health. So the person with the single biggest influence over whether someone comes to work on Monday has usually been given no help at all in doing it.

Gallup’s numbers make the same point from a different direction. Compare business units in the top quartile for engagement against the bottom quartile and you get 81% lower absenteeism, 64% fewer safety incidents, 41% fewer quality defects, 18% higher productivity and 23% higher profitability. Those are median differences across a vast sample. In a sector where safety incidents and defects carry real financial weight, that’s not a wellbeing statistic. That’s an operations statistic.

Turnover and lost knowledge. Make UK’s labour turnover research showed sector turnover falling to 10.85% in 2024, down from 20.75% in 2022, with voluntary turnover at just over 6% and retirement now the most common reason for leaving. On the face of it, good news.

I’d be cautious about celebrating, though. A cooling labour market keeps people in place whether they’re happy or not. And an ageing workforce means the departures you’re seeing are the ones carrying thirty years of undocumented knowledge out of the door with them. Make UK’s Skills Commission identified 55,000 long term unfilled vacancies in the sector, costing an estimated £6 billion in lost output a year, with 42% of manufacturing vacancies now classed as skills shortage vacancies, up from 29% in 2017. Around 36% of the workforce is over 50.

Replacement isn’t cheap either. CIPD research puts it at 75% to 200% of annual salary depending on the role. Oxford Economics landed on an average of £30,614 per employee. And an unfilled CNC machinist post on a three-shift operation has been estimated at £8,000 to £15,000 a month in lost output and overtime.

A formula you can do on the back of a delivery note

Here’s the calculation. It’s deliberately crude, because a crude number you actually finish beats an elegant model you abandon.

Annual cost of disconnection = A + B + C + D

A. Missed deadlines and disruption. Take the hours of stoppage or delay per month that you’d honestly attribute to coordination failures rather than genuine equipment failure. Multiply by twelve. Multiply that by your contribution per productive hour. Then add expedited freight, recovery overtime and any customer credits from the last year.

B. Preventable mistakes. Take your annual scrap and rework value. Estimate the share caused by unclear instructions, handover gaps or shift variation rather than material or machine faults. A third is a common, conservative starting point.

C. Morale and absence. Take your absence days per employee, subtract a realistic benchmark, multiply the excess by headcount, then by the fully loaded daily cost of covering that absence.

D. Turnover and lost knowledge. Take last year’s leavers, strip out retirements and genuine life events, and multiply what’s left by a conservative replacement cost.

Let me run it for that West Midlands firm. Roughly 140 people, £18 million turnover, precision engineering.

A. They reckoned about six hours a month of stoppage or delay came down to coordination. Wrong tooling staged, wrong revision, maintenance not informed. That’s 72 hours a year. Their contribution per productive hour worked out at roughly £1,285. So £92,520. Add £25,000 for expedited freight and recovery overtime across the year. Call it £117,000.

B. Scrap and rework at 1.5% of £18 million is £270,000. Attribute a third to communication and handover issues. £90,000.

C. Absence running at 10 days against a realistic target of 6. Four excess days across 140 people is 560 days. At £180 a day fully loaded, including agency premium and lost output. £100,800.

D. Around 15 leavers. Seven were retirements or genuine moves. Eight were avoidable. At £15,000 each, which is half the Oxford Economics average and deliberately cautious. £120,000.

Total: £427,800 a year. About 2.4% of turnover.

Now the return. You will not fix all of that, and you shouldn’t claim you will. Assume you recover 20%, which is modest against Gallup’s quartile differences. That’s £85,560 a year.

Say the investment is £40,000. Supervisor development for your team leaders, a proper structured handover process, and some facilitated cross functional problem solving.

ROI = (£85,560 minus £40,000) divided by £40,000 = 114%. Payback in roughly five and a half months.

And here’s the bit I’d hold in reserve. Halve every single one of those assumptions. Recover only 10%. The thing still pays for itself inside the year.

Taking the number upstairs

You’ll get three objections. I’d have them ready.

“Those numbers are made up.” They’re estimates, and you should say so first, before anyone else does. Then point out that the capital expenditure case for the machine you approved last year was also built on estimated utilisation and estimated yield improvement. Everyone was comfortable with that. The difference here isn’t rigour, it’s familiarity. Offer to show the calculation with the assumptions halved. If it still works, and it will, the argument is basically over.

“We can’t attribute that to communication.” Fair challenge, and the honest answer is you can’t attribute all of it. So don’t try. Pick the incidents where the root cause analysis already says what you need it to say. Most manufacturers have months of nonconformance reports with root causes like “operator not informed of revision change” or “handover not completed.” That’s your evidence, and you didn’t have to generate it. It’s been sitting in your quality system all along.

“This is a soft skills thing.” This is where the Gallup data earns its keep, because those are hard operational outcomes. Fewer defects. Fewer safety incidents. Lower absence. Higher productivity. Nobody in a manufacturing business thinks safety incidents are soft.

One more piece of advice, and it’s the one I’d actually lead with. Don’t ask for the whole thing. Ask for one line, one shift pattern, one supervisor group, for one quarter, with agreed measures taken before and after. A pilot is much easier to say yes to than a programme, and a pilot that works builds a business case nobody can argue with, because it’s built from your own plant rather than somebody else’s research.

What actually shifts the number

Briefly, because this could be its own article.

Start with your supervisors. It’s the highest leverage thing available to you, and it’s usually the most neglected. The people running your shifts were promoted because they were excellent at the job and then handed responsibility for fourteen humans with no preparation. Give them the basics. How to run a decent conversation. How to give feedback that doesn’t land as an attack. How to escalate a concern without it becoming a row.

Fix your handovers properly. Not a form. A short, structured conversation with a consistent shape, at a consistent time, with a consistent set of things covered. It sounds almost too simple to be worth the effort. It’s usually the single cheapest improvement available in a manufacturing business.

Put the functions that blame each other in the same room on a regular basis, working on something concrete. Maintenance and production. Quality and operations. Not a workshop with flip charts. An actual recurring problem with an actual owner and an actual deadline.

And measure it. Absence, scrap by root cause, delivery performance, near misses, exit reasons. Take a baseline before you start, because the most common way these initiatives die is that they work and nobody can prove it.

The cost is already being paid

I’d finish on this.

The money in your four buckets isn’t hypothetical, and it isn’t a future risk you might one day face. You paid it last year. You’re paying it this month. It’s leaving the business right now in six-hour chunks of coordination failure, in scrap that traces back to a conversation that didn’t happen, in the four extra sick days that were really about a supervisor who never learned how to have a difficult conversation, and in the retiring toolmaker whose knowledge nobody wrote down.

The only genuine question is whether you’re counting it.

Because once you count it, the conversation changes completely. You stop asking permission to invest in your people and start explaining why the business is choosing to absorb a six-figure annual loss it could reduce for a fraction of the cost. That’s not a training request. That’s an operational improvement proposal, and it belongs in the same queue as the capital expenditure it’s currently losing out to.

Do the sum. Use your own numbers, be deliberately conservative, write down every assumption so nobody can accuse you of hiding one. It’ll take you an afternoon and a couple of conversations with your finance and quality people.

I’d be surprised if the number comes back small. It very rarely does.

Ready to put a number against the fix?

Here’s the slightly awkward bit about the sum you’ve just done. The moment you’ve got the figure, doing nothing becomes a decision rather than an oversight.

So if the number came back bigger than you were expecting, these are the two levers I’d pull first, and they map almost exactly onto the two problems in this article.

Helping Managers to Succeed and Lead is for that supervisor layer. The people who got promoted because they were brilliant at the job and then had fourteen humans handed to them with no preparation. It pairs each manager with a dedicated mentor, starts with the Igniting The Leader In You training, and then follows a 90-day milestone structure with reviews at 10, 30, 60 and 90 days. That last part matters more than it sounds. Most training fails not because the content was wrong but because nobody came back three weeks later to ask how it was going.

High Performing Teams in the Workplace is for the other half. The maintenance lead who stopped speaking up. The two shifts running the same cell differently. It’s a full day workshop, on site at your facility, for up to twelve people, built around five pillars including a working rhythm for solving problems and the trust to challenge each other without it turning into a row. It’s £1,400 per group, excluding VAT and travel, which is under £120 a head for a team of twelve.

Put that against the number you calculated earlier. I’d guess the arithmetic makes itself.

If you’d rather just talk it through first, that’s genuinely fine. Call 0330 311 2820 or start with a fifteen-minute conversation about where your teams are now and where you need them to be. No pitch. Just a look at your figures and an honest view on what’s worth fixing first.