Cut headcount. Book the savings. Tell the board. Celebrate the stock price bump.
That's the AI playbook in most companies right now. Stand up a chatbot or agent. Eliminate the team that used to do the work. Move the savings into next quarter's earnings call. Repeat with the next function.
It's a small idea dressed up as a transformation strategy. And the companies running it are about to get out-strategised by the ones that aren't.
IKEA went the other way. Worth understanding what they saw that everyone else missed.
What IKEA did
In 2021, Ingka Group — the IKEA retail arm, operating across more than 30 countries — built a customer service chatbot. They called it Billie, a nod to the iconic Billy bookcase. Billie handles tier-one inquiries: order tracking, returns, product questions, the routine stuff. About 47% of customer interactions get resolved without ever reaching a human.
So far, this is a familiar story. AI does the boring work. Headcount comes out. Big productivity number lands in the board deck.
That's not what IKEA did.
Instead of laying off the customer service team whose work had been automated, IKEA looked at the other 53% — the inquiries Billie couldn't handle. And here's the part that matters, because it's the part you can copy: they didn't count those tickets. They read them.
When they read them, a pattern jumped out. A lot of customers weren't asking about an order. They were asking for help designing their home. How do I make this sofa work in my flat? What do I do with this awkward corner? Which lamp goes with that table? Real interior design questions that Billie was never going to answer well.
So IKEA reskilled 8,500 call-centre workers into interior design advisors, and spun up a paid remote design service powered by these newly trained humans.
In its first year, that new service generated €1.3 billion in revenue — 3.3% of Ingka's total. They're now targeting 10% by 2028.
Read that again.
The 8,500 people the standard playbook would have cut became the seed of a new billion-euro business line. Not by accident. IKEA treated AI as a way to free people up to do something more valuable, then did the unglamorous work of figuring out what that something was — by reading the demand that was sitting in their own support queue.
Ulrika Biesert, IKEA's Chief People & Culture Officer, put the principle plainly: "People have been at the heart of IKEA for over 80 years — and that's exactly where they'll stay."
Read that as a strategic commitment, not a sentimental one. It rules out a whole class of AI moves and forces a better question.
Why this isn't an outlier
There's a name for this pattern, and it's almost two centuries old.
In 1865, the economist William Stanley Jevons noticed something strange about the steam engine. James Watt had made it dramatically more efficient. Engines now needed far less coal to do the same work. The conventional wisdom was that Britain's coal consumption would plummet.
It exploded.
Cheaper, more efficient engines made steam viable in industries where it had previously been too expensive. Factories used more of them. New use cases emerged. Total coal consumption went up, not down.
That's the Jevons paradox: when you make a resource more productive, you don't reduce demand for it. You expand the things you can do with it.
The same pattern shows up everywhere. Spreadsheets didn't reduce demand for analysis — they created industries built on it. Cheap web hosting didn't shrink the internet; it made everyone a publisher. Better compilers didn't put programmers out of work; they produced more software than humans could keep up with.
AI is a Jevons machine. Every routine task it absorbs lowers the cost of customer service, of design work, of everything in the white-collar workflow. Lower cost means new things become possible — things that were too expensive to attempt before. The ROI shows up in what you can suddenly afford to do.
In my leadership coaching, we end up talking about leverage a lot. What's the highest-leverage part of your role, and how do we make more space for it? AI is the biggest lever we've ever seen — and it's leveraging entire organisations, not just leaders.
IKEA grasped this. A free hour of a customer service rep's time isn't a cost saved. It's an hour you can spend doing something the company couldn't previously afford. The question stops being how many heads can I take out? It becomes what couldn't we do before that we can do now?
Two completely different questions. Two completely different companies.
Why most execs are getting this wrong
Cost-out is comfortable. It's measurable. It hits the next quarter. CFOs love it. Boards understand it. You can build a slide for it.
And it isn't wrong. Savings are real, but the mistake is stopping there. Cost-out is the floor. The companies treating it as the ceiling are handing the upside to whoever is willing to think bigger.
Because growth requires something harder. It requires you to actually have a point of view. About where your business is going. About what your customers are trying to do that you don't yet help them with. About what you'd build if your team had capacity. About whether your strategy is still the right one in a world where the cost curve of human work just collapsed. And then using those savings to fund these bets.
That's hard. So most companies skip it and do the easy thing.
The easy thing is a trap. If your only AI strategy is to do what you already do with fewer people, your competitor's AI savings won't show up as cost reductions on their P&L. They'll show up as new products, new services, new revenue streams that you don't have an answer to.
By the time you notice, the gap is structural.
Going up the stack
I write a lot about the Decision Stack, a mental model for how decisions connect — from vision and mission at the top, down through strategy, objectives, opportunities, to the principles that guide daily choices, and back up again.
Most AI conversations I see in companies are stuck at the bottom two layers. Which tasks should we automate? Which workflows should we redesign? Which tools should we buy? Useful questions. You can answer all of them perfectly and still be doing the wrong thing.
The IKEA question is a Strategy question. Maybe a Vision question. What is AI freeing us up to become? You don't get there by optimising more workflows. You get there by stepping back from the workflows entirely and asking what your business is for. What you'd do for your customers if cost wasn't the constraint. Where the next billion of revenue might actually come from.
Biesert's quote about people being at the heart of IKEA is a strategic boundary condition. It rules out a whole class of AI moves and opens up another. That's what going up the stack looks like.
The real question
If you're a CEO, COO, or board member right now, the question your AI strategy needs to answer isn't how much can we save? It's what can we now do that we couldn't do before?
The companies that get that right won't be the ones with the leanest org charts. They'll be the ones with the new revenue streams nobody saw coming.
And the first place to look isn't a strategy offsite. It's the demand already sitting in front of you, in the questions your customers are asking that nobody has time to answer well.
IKEA found a billion of it in their unresolved tickets. What's hiding in yours?
