Four years of apprenticeship, and on the day he finished it, he was gone. Somebody down the road waved a lot more money at him and he took it, which is what most of us would have done at that age with a fresh ticket in our hand and the ink still wet on it.

He didn’t last at the new place. The people who’d poached him worked it out almost straight away, and he ended up back with the employer who’d trained him, on less than he’d walked out for.

He was qualified. He wasn’t competent.

A bloke told me that in Brisbane, about his own apprentice, in a room full of business owners. The story wasn’t what stayed with me. What stayed with me was that nobody in the room needed it explained.

I’ve told it twice since. Once in a workshop I ran for emerging leaders at a Melbourne TAFE, which is to say the people whose institution issues the qualification, and eight days later from a stage on the Gold Coast in front of a room of civil contractors, which is to say the people who pay for it, carry it, and lose the person at the end of it. Both rooms nodded before I got to the end of it.

When the employer who trained him, the institute that certified him and the industry that hires him all recognise the same story, it isn’t an anecdote any more. And the number sitting underneath it, apprenticeship commencements in Australia, has been going the wrong way for three years while almost nobody puts it next to the number that makes the whole system look healthy.

The bit I didn’t expect to run longest

The first of those rooms was a three-hour workshop I ran for emerging leaders at Holmesglen TAFE in Melbourne, and they’d come from right across the institute. Teaching faculties, health and community, trades, student services, language, HR, all of them in the one room arguing with each other, which is generally where the good stuff is.

I started teaching there in 1990, back when the building we were sitting in didn’t exist.

The session was built around leading people into a future that hasn’t been built yet, and whether you can make it one they’d actually want to live inside, which is the whole of Inhabitable Futures in a sentence. I’d assumed the block that would run away with the morning was the AI one, because it almost always is.

It wasn’t. What took the morning over was a much older question, and they took it off me and ran with it.

What is a qualification actually worth now?

One of the trades staff pushed back at me, firmly and correctly, that in his world the regulation makes the qualification non-negotiable and you can’t do what he does no matter how many YouTube videos you sit through. Somebody else asked whether financial literacy sits anywhere in the building curriculum, and that produced a properly useful answer about estimating, costing and legal risk that half the room hadn’t known was in there.

A room of educators, arguing about the worth of the thing they hand out. Not one of them was talking about technology.

Somewhere in the middle of it I told them about the apprentice.

Then again, to the other side of the transaction

Eight days later I was giving the keynote at the CCF Queensland state conference on the Gold Coast, in front of a room of civil contractors. Different sector, different state, a completely different set of problems.

First question in the Q&A came straight at me. Apprenticeships. Four years. Do you see that getting whittled down, and does it need to be, to get the workforce moving?

Out came the same story, unplanned.

At the TAFE I’d told it to the people who issue the qualification. On the Gold Coast I told it to the people who lose the person at the end of it, and what landed in both rooms was the same thing. The certificate was real and so were the hours, and neither of them was the point. The employer who’d spent four years with him had him priced accurately the whole way through. The one who’d never met him priced the certificate.

Apprenticeship commencements in Australia, and the number nobody puts beside them

Two figures, same agency, same dataset, moving in opposite directions, and the gap between them is where the next decade is hiding.

The National Centre for Vocational Education and Training publishes the numbers on this, and its most recent data covers the twelve months to 31 December 2025.

In that period, trade commencements in Australia fell 4.2 per cent, down 3,390, to 76,585. That’s the third consecutive annual decline.

In the same period, trade completions hit record highs for electrotechnology and telecommunications trades workers, with electricians among the strongest growth.

Read those two sentences again. Record completions. Falling commencements. Third year in a row.

Both are accurate. Both come from the same dataset. And almost every piece of commentary I’ve seen quotes one of them and not the other.

If you only read the completions line, the system looks like it’s working. Record numbers of electricians finishing. Good news, move on.

If you only read the commencements line, the system looks like it’s collapsing.

Neither reading is right, because the two numbers aren’t describing the same thing. They’re describing the same pipeline, measured at opposite ends, four years apart.

The good number is a rear-view mirror

A completion in 2025 is a person who started in 2021 or 2022. It’s a report on a decision somebody made four years ago. It tells you what an employer was willing to commit to, what the incentives looked like, and what the economy felt like, back then.

It is genuinely good news. It’s just news about the past.

A commencement is the other thing entirely. A trade commencement today is a qualified tradesperson in 2029 or 2030, assuming they finish, and the six-year completion rate for electrotechnology and telecommunications trades workers sits at 68.3 per cent for the cohort that started in 2019. So roughly two in three.

Which means the commencements line is not a statistic about now. It’s the only workforce number on the board that’s actually about 2030.

And 2030 is a crowded year in this country. It’s the year almost every energy target lands, and it sits just past the window the housing accord is measured in. The Powering Skills Organisation puts the requirement at roughly 42,000 more licensed electricians by 2030, plus something in the order of 30,000 more energy trades workers on top of that. That is a lot of people who need to have started an apprenticeship by about now.

They haven’t. That’s the whole point. Commencements have fallen for three years running.

This is what I mean when I talk about Immediate Futures. Not what’s coming in ten years. What’s already here, sitting in a spreadsheet, needing attention today because the lag is built into the physics of the thing. You can’t decide in 2029 to have more electricians in 2030. That decision closed a while ago and hardly anybody noticed it closing.

What I said in 2015, and the condition nobody met

I’ve been talking about this since 1990, when I was the one standing at the front of the classroom, and it isn’t a subject I visit occasionally. It comes up most days, in most rooms, in sectors that have nothing to do with education, because who gets taught what and by whom sits underneath almost every other question a leader asks me. I was on 6PR in 2010 running through the jobs of the future, and that was already twenty years into it.

In May 2015 I did one of my regular segments with Phil Staley on ABC Far North Queensland, and the question that day was whether there would still be jobs in Cairns in 2025. I argued that routine work was going wherever you looked, that new careers would arrive in its place, that being in a regional town would stop being the barrier it had always been, and that people would end up managing their own working lives instead of waiting for an employer to hand them the next rung.

That much landed. The routine work went, the new careers arrived, and distance stopped mattering the way it used to.

But there was a condition at the end of that piece, and it’s the condition I’d underline now:

“to take advantage of these new horizons we will need to cut the anchor with many of our long-held beliefs and work models and we will need to evolve our education system whilst simultaneously up-skilling and providing confidence to the workforce of today.”

2025 came and went. The jobs half of that arrived more or less as described. The education half didn’t, and the NCVER numbers show you exactly why not: a training system can’t turn quickly, because the thing it produces takes four years to build, which means the response to any signal you notice today doesn’t show up until 2030, and by then you’re reading a decision somebody made while you were still noticing.

I came back to it in October 2024, in a piece on the future of apprenticeships in a post-automation world, arguing that apprenticeship as an idea would become more important rather than less, because the specialised, hands-on, judgement-heavy work is precisely what automation doesn’t take. I’d still argue it. The idea being right doesn’t rescue the structure, though. The notion of apprenticeship is imperative. The four-year container it arrives in is a different question, and it’s the container that’s under strain.

The bottom rung and the middle rung are going at the same time

This is the part I gave the civil contractors, and it’s a Ripple Effects read rather than a headcount read.

The first-order story is the obvious one. There aren’t enough tradespeople and we need more.

The second-order story is the difficult one, and it has two parts running at the same time.

The bottom rung. A great deal of the routine work that apprentices used to be employed to do is being done by machines and by software. Not wholesale, and not overnight. But it only moves in one direction, and each year there’s a little less of it left. That work was never really about the work. It was the reason you had a junior on site at all, and it was the scaffolding underneath which somebody slowly became competent. Take away the economic reason to have a first-year on the tools and you don’t just lose a task. You lose the entry point. I wrote about this happening at the other end of the collar, in white-collar work, in We Didn’t Just Automate the Jobs. We Removed the Years. The mechanism on a work site is identical, and it’s further along.

The middle rung. Middle management is going too, through age, through restructuring, through the automation of the coordinating work middle managers used to do. And middle management, in almost every trade and profession I look at, is the rung that traditionally brought the apprentice in. Taught them. Corrected them. Told them off. Stood next to them while they made the mistakes that turn a qualification into a capability.

Those two rungs used to work together, and pulling both out at once gets you precisely the apprentice that Brisbane room recognised. Four years of hours. A certificate at the end of them. And nobody standing beside him for the part that was never going to be examinable.

The question I put to the TAFE room, and I’ll put it here. Where does anybody now learn to go from competent to great? Where do they make their mistakes? And who’s going to be next to them, willingly or grudgingly, to show them, tell them, do it with them?

If your organisation can’t answer that with a name and a mechanism, you don’t have a development pathway. You have a hope.

Four years isn’t the answer either

There’s a lazy conclusion sitting right next to this one and it’s wrong, so let me put it out of the way. I’m not arguing the four years should be shorter. The trades bloke who pushed back at me in that TAFE room was right. Regulation makes the qualification non-negotiable in his world and it should, because there are things you can’t learn in six months and there are consequences to getting them wrong that don’t apply to most other work.

But look again at what that second employer found out, and how fast he found it out. Four years of elapsed time didn’t produce competence either. It produced a certificate.

So the useful question was never how long the pathway ought to be. It’s which parts of it are doing the work and which parts are there because that’s how long it’s always been, and as long as the completion numbers kept looking healthy nobody had much reason to go and check.

What this actually means if you employ people

Naming a pattern is the cheap part, so here’s where it bites.

Your workforce reporting is almost certainly backwards. Most organisations I work with report completions, retention, graduate numbers and attrition. Every one of those is a lagging indicator. They’re all reports on decisions already made. If your board pack has no leading indicator in it, your board is steering by the wake.

The equivalent question in your organisation is not how many people finished the programme. It’s how many started one this quarter, how that compares with the last three, and what capability that implies four years out.

The person who trains is disappearing before the person being trained. If you’re cutting middle management, you’re almost certainly not counting the teaching that layer was doing off the books. It never appeared in a position description. It won’t appear in the savings calculation either. It’ll appear in about three years, as a capability gap nobody can trace back to a decision.

Poaching is a symptom, not a strategy problem. When trained people are scarce, buying one is always cheaper than making one, right up until nobody is making any. The firms carrying the training cost are the ones least able to carry it. That’s not a moral observation, it’s a structural one, and it’s the shape of every industry where the cost of building capability sits with one party and the benefit sits with another.

Qualified and competent are now two separate purchases. If you hire on the certificate alone you’re buying a proxy that used to be reliable and isn’t any more. Somebody in your organisation has to be able to tell the difference, and that somebody usually sits on the rung you’re in the middle of removing. This is the same question HUMAND asks about machines, pointed at a person: not what the label says the job is, but which tasks actually sit inside it and who can genuinely do them.

Four things I’d do about it

Four of them, and none of them costs much beyond attention.

Put a leading indicator in the pack. One number, reported quarterly, that describes intake rather than output. Apprentices started. Juniors hired. People entering the development pathway. Track it against the same quarter last year. It’ll be a smaller, less flattering number than the one you report now, and it’ll be the only one telling you anything about 2030.

Name the person who teaches. Not a mentoring programme. A person, a name, and time in their week that is protected and paid for. The moment teaching is something people do in addition to their real job, it becomes the thing that goes when the quarter gets tight.

Separate what the four years is buying. Go through the pathway and mark each part as either genuinely time-dependent, or historically time-dependent. Some of it is exposure that can’t be compressed, because you have to see enough situations. Some of it is sequence that could be re-cut. You won’t know until you look, and almost nobody has looked.

Assume the person you train will leave, and train them anyway. The alternative is a sector where nobody trains anyone and everybody bids for the same shrinking pool. If you want the mechanism that makes it survivable, it’s the guild answer, and it’s older than any of us. The obligation to train is shared across the trade rather than carried by whoever happens to employ the person on the day.

Where I’ve landed

One business owner’s story, carried into a workshop full of educators and then onto a stage in front of a room of contractors, and every one of those rooms nodded before I got to the end of it.

Which tells me the problem isn’t hidden at all. Everybody in the system can see it perfectly well from where they happen to be standing. What nobody has is the shape of the whole thing, because each of them only ever sees their own end of a four-year pipe.

The institute sees completions holding up and the credential being questioned anyway. The employer sees a trained person walking out the door the week he qualifies. The policy conversation sees a shortfall number for 2030 and reaches for incentives. Every one of them is looking at something real. Not one of them is looking at the same thing.

Two numbers. Record completions, falling commencements. One of them is a photograph of 2021. The other is a forecast for 2030, and it has been going the wrong way for three years while we’ve been reading the photograph.

The class of 2030 hasn’t started. That’s not a prediction. It’s arithmetic, and it’s already on the record.

Choose Forward.

Frequently asked questions