You Can't Hire Your Way Out of This One

UK construction needs 41,200 extra workers a year until 2030. There’s a skills shortage that is becoming increasingly untenable. 

Here’s what that means for the future of field service businesses, and why the first thing that’s limiting growth isn’t actually the number of engineers themselves.

How bad is the trades labour shortage?

I worked in vocational education for 3 years at the start of my career. I began as an apprentice working with the trade industry and spent time on-site with apprentices working towards the same qualifications. The signs of an impending crisis were glaringly obvious over a decade ago – but even with hindsight, it still feels staggering to see where we are today.

Recently, it’s been cited that the UK construction industry needs 41,200 extra workers every year between now and 2030. That's more than 206,000 people in five years, just to keep up with work that's already been forecast. Look further out and the figure is something like 1.3 million new skilled tradespeople, plus 350,000 apprentices, over the next decade.

Now, on the other side of the coin: retirements are running well ahead of replacements. 

In the US,(where the data is better but the story is identical), roughly 600,000 skilled trades jobs were posted last year against about 150,000 new workers coming through apprenticeships. 

Put simply: for every five people who retire, only two arrive.  Around 22% of construction workers are already 55 or older. 

‍You don't need any of that to know the problem, mind. You’ll know it from the last time you tried to fill an engineer vacancy, and from what you ended up paying to close it. 

Nearly three quarters of SME builders say a lack of skilled tradespeople is affecting them. Half have had jobs delayed. More than a fifth have cancelled work outright. In the commercial world, the share of contractors reporting rising wages went from 55% to 71% in a single year.

Here's the bit worth sitting with, though. This isn't a cycle. Cycles turn. A gap this size (in a trade where competence takes four years to build and can't be shipped in from somewhere else), either closes over a decade,or it doesn't close at all.

So, any plan that starts with "once we've got a couple more engineers in" is a plan competing against a national shortage. And the national shortage is winning.

Which makes the useful question a different one completely.

It now becomes: how do you hire faster than everyone else? What can you grow without hiring?

What's really capping your growth?

Say you've got 30 engineers. That's a fixed amount of field capacity. 

What decides your revenue is how much of it turns into completed, invoiced work, and that conversion happens in the office, not the van. 

Think about the enquiry that rings out at six on a Friday. The quote you sent three weeks ago and never chased. The customer who forgot the appointment, so your engineer spends forty minutes driving to a locked door.  The job sheet that lands three days late, so invoicing slips a week and the cash slips with it. Every single one of those is capacity you've already paid for and subsequently put in the bin.

It's also why adding engineers so often doesn't add the revenue you expected. 

Put two more engineers on schedule and you haven't fixed it, you've made it heavier, because every engineer brings more scheduling, more paperwork, more customer calls and more invoicing for the same coordinator to absorb. 

How can AI actually help here? 

This is where most AI-in-the-trades conversations fall over. They start in the field: the robot plumber, the self-driving van, the gadget that replaces twenty years of knowing what that noise means. Everyone sensibly decides it's a decade off, and stops listening.

The biggest study we have on how people actually use AI at work, built on 200,000 real conversations, found the same thing over and over. 

What AI does well is information work: creating it, processing it, passing it on. The jobs that scored lowest were nearly all the physical ones. Construction trades, plant operation, grounds maintenance. The work that needs hands, machinery, and being in the room.

If you wanted reassurance that nothing's coming for your engineers, there it is: from those  building the models.

But turn it round and it gets more interesting.. The work AI is good at is language and coordination. In a field services business, none of that happens in the van. All of it happens in the office.

Answering the phone and working out what the caller actually needs. Asking the qualifying questions. Finding the right slot for the right skill set. Sending the confirmation. Chasing the quote. Nudging the customer the morning before. Turning what the engineer says into a job record.

That's the list. And it's exactly what technology can already do. 

Now, not in three years.

So the question was never "can AI replace an engineer". It's "how much of the work stopping my engineers being productive is information work?" Answer that properly and the number tends to be uncomfortable.

What does this look like in practice?

The practical version of this is deeply unglamorous, which is usually a good sign.

Enquiries stop leaking. Every call gets picked up, at 2am, on a bank holiday, by something that can qualify the job and book it rather than take a message. For a lot of firms this is the single biggest recoverable line in the business, because the out-of-hours calls are the ones with the highest intent. Nobody with no heating leaves a voicemail and waits. They ring the next firm on the list.

Quotes get chased, every time. Not when someone remembers. Not when the pipeline goes quiet. The gap between converting 40% of your quotes and converting 80% usually isn't a better quote. It's the third follow-up nobody has time for.

Engineers stop driving to locked doors. A confirmation the day before is a two-minute job that never gets done, and no-shows are the most expensive avoidable cost in field services. You pay for the whole visit and get nothing back.

Paperwork stops being a tax on the engineer. Fifteen minutes a day, times 30 engineers, is roughly a full-time role's worth of time. Spent on data entry. By the most expensive and hardest-to-replace people you employ.

None of that is transformation. It's the same business, doing the same work, with the leaks closed. What you get out of it is capacity: more finished jobs from the engineers you've already got, and a way to grow that doesn't depend on winning a hiring race against your whole industry.

The industry is getting there, and fast. 

Recent research across the trades found 74% of residential contractors now see AI as a way to run more efficiently, and 73% think moving early is a competitive advantage. 

But a tiny 25% are actually using it. 

Among commercial contractors, the share reporting real, measurable impact more than doubled in a year, from 17% to 38%. Of the firms that have gone ahead, 62% report measurable gains, and plenty are saving three hours a week or more.

That gap between what people believe and what they've done is the interesting part. It won't stay open long.

How do you get this right?

Two things worth being clear about, because they change how you sequence the work.

Keep investing in the field. None of this trains an apprentice, and the trades still need tens of thousands of them. If anything, the case for apprenticeships gets stronger: when engineer time is the scarcest thing in your business, protecting it and growing more of it is the whole game. AI gets you more out of every engineer you have, which is exactly why the ones you have are worth more.

Plan the rollout, not just the purchase. This is where results are won. When contractors were asked what was actually holding them up, the answers were training (44%), integration being a pain (44%), not really understanding the tools (38%) and not knowing what the return looks like (37%). Staff resistance came last, at 18%, so the people doing the work are already with you. Budget a fortnight for training and integration and you have solved most of the failure modes in the industry.

Does it read and write to the platform you already run on, so the job record is right without anyone typing it in twice? 

Does it hand over to a human when it's out of its depth, or does it quietly get it wrong and lose you a customer? Can you see what it did, and why? 

And are you buying an outcome you can count, like jobs booked, quotes converted, admin hours handed back, or are you buying a licence and a hope?

The bar is trust at scale, and trust gets earned in the unglamorous details: the handover, the audit trail, the integration that holds up on a Monday morning. Get those right and the rest follows quickly.

Where does that leave you?

The labour shortage is the defining commercial fact of the next ten years in the trades, and you can't hire your way out of it. For firms that only know how to grow by adding people, that's bad news.

But the constraint cuts both ways. Everyone is short of engineers. Very few firms have worked out that the office is where their engineers' capacity is quietly going missing, and fewer still have done anything about it. That asymmetry won't last, and while it does, it belongs to whoever moves first.

Start by measuring the leaks instead of guessing at them. How many enquiries went unanswered last month? What share of quotes got a second chase? How many visits were no-shows? How many engineer-hours went on paperwork?

Frequently asked questions

How many extra workers does UK construction need?

41,200 a year between now and 2030, or more than 206,000 over five years. The longer-run figure is around 1.3 million new skilled tradespeople plus 350,000 apprentices over the next decade.

Will AI replace tradespeople?

No. The largest study of how AI actually gets used at work found the lowest applicability scores in manual occupations, including construction trades and plant operation, because that work needs hands, machinery and being in the room. What AI is good at is the information work in the office.

Can AI solve the skilled trades shortage?

Not directly, but it changes what the shortage costs you. It gets more completed jobs out of the engineers you already have by closing the gaps between an enquiry arriving and an invoice going out.

Why doesn't hiring more engineers increase revenue proportionally?

Because the office is usually the constraint, not the field. Every engineer you add brings more scheduling, paperwork, customer calls and invoicing for the same coordinator to absorb, so field capacity rises while the ability to convert it into invoiced work doesn't.

What is stopping contractors adopting AI?

Training (44%), integration complexity (44%), not understanding the tools (38%) and unclear return (37%). Staff resistance is the smallest factor at 18%.

Where should a trades business start?

By measuring the leaks: unanswered enquiries last month, the share of quotes that got a second chase, no-show visits, and engineer hours spent on paperwork. Most firms have never put a number on any of it.

Elyos builds AI agents for field services businesses: voice, email and operational intelligence, integrated with Simpro, ServiceTitan, JobLogic and Commusoft. The agents answer the calls, chase the quotes, confirm the appointments and get the job notes written up. One of our customers has given their engineers around two hours a day back on paperwork alone. If you'd like to see what that looks like against your own numbers, get in touch.

‍