IKEA’s AI chatbot saved about €13 million in call center costs. Retraining the 8,500 people it freed up helped build a €1.25 billion sales channel.
Yet every time I talk AI with executive teams, the conversation still kicks off with the smaller number. It’s like everyone’s hypnotized by the cost savings, while the real story is hiding in plain sight.
So let’s talk about Billie, the Swedish meatball of chatbots. IKEA’s story might just be the best AI workforce case study your boardroom has never heard of, unless someone’s been sneaking in some light reading between quarterly reports.
In 2021, Ingka Group rolled out an AI customer service assistant named Billie. Ingka is the largest IKEA franchisee and operates in 31 countries. And yes, Billie is named after the Billy bookcase. Somewhere in Sweden, a branding team high-fived.
Billie took the routine stuff: store hours, order status, stock checks, and return policies. Over its first two years, it resolved about 47% of conversations with the customers who used it. Fortune reports that number is now up to 74%!
This is where the standard playbook says, "Great, time to cut." You automate nearly half your call center’s work, so you trim nearly half of it. The spreadsheet smiles, and the earnings call is a parade of high-fives.
Over roughly two years, it retrained about 8,500 call center workers. Some moved up to handle the complex service issues Billie couldn’t touch. Many more became remote interior design advisors. Now they help customers plan kitchens, untangle awkward room layouts, and, most importantly, buy things.
By IKEA’s own numbers, yes. (Fair warning: these are company-reported figures, not an independent audit.)
IKEA now has 24 remote sales centers covering all 31 Ingka markets. For three years running, they’ve been the company’s fastest-growing sales channel, up 15% to 20% a year. Revenue hit €1.25 billion (roughly $1.37 billion) last fiscal year, up from €1.08 billion the year before.
|
|
Automating the tasks |
Redeploying the people |
|
What it did |
Billie resolved 47% of conversations (now 74%) |
About 8,500 workers retrained as design advisors and complex-case specialists |
|
Reported value |
About €13 million in call center savings |
€1.25 billion in remote sales revenue last fiscal year |
|
Growth |
One-time cost reduction |
Fastest-growing sales channel, up 15% to 20% a year for three years |
Even if you give the retrained workforce credit for just a fraction of that growth, it’s not even a contest. Redeploying the people paid off far more than automating the tasks did.
That’s the part that should make every CPO sit up, spill their coffee, and start asking different questions. It also showcases, to the tune of billions of dollars, that AI transformation is a people challenge.
Cutting is easy to model. Redeployment? That’s where the spreadsheets start sweating.
A headcount reduction is a tidy line item for next quarter. Redeployment is messier. You need to know, in detail, what your people actually do and what skills they bring to the table. You also need to know where else in the business those skills could move the needle. Most organizations can’t answer those questions. They know job titles, not the work hiding behind them.
And the anxiety on the ground is real. Mercer’s Global Talent Trends 2026 report found that 40% of workers now fear AI will make their jobs obsolete, up from 28% in 2024. When leaders reach for layoffs as the default AI story, that fear becomes self-fulfilling.
It isn’t always the company’s fault. They may try to lead with a strong people-and-trust focus, but the media and LinkedIn news sidebar still showcase layoffs. The neuroscience kicks in, and then folks go into fight, flight, or freeze. Plus, the people you most want to keep are the first ones polishing their LinkedIn profiles.
This is the gap where I spend most of my time at livingHR. When we sit down with leaders about AI, the first question is almost never “which tool?” It’s “what work is actually happening here, and what happens to it once the tool shows up?” If you can’t answer that second part, you’re not doing workforce strategy. You’re just doing cost-cutting with a fancier label.
A few things stand out, and none of them are about the chatbot.
They treated the chatbot leftovers as a signal. The share of queries Billie couldn’t resolve wasn’t a failure rate. It was a demand signal. In those conversations, customers wanted judgment, taste, and a human who could talk them out of a kitchen island that will never fit. IKEA followed the demand instead of the savings.
They mapped skills to new value, not old roles. A call center agent already knows how to listen, de-escalate, and solve problems under time pressure. Those are sales and design consulting skills wearing a headset. IKEA’s training covered room-planning tools, measurements and color preferences. It also taught the probing questions that uncover what a customer actually wants. One advisor in Sheffield went through a two-week “kitchen school” plus store field trips. New hires now get five to six weeks of training.
They gave it time. Two years. Not a pilot, not a lunch-and-learn, not a Coursera license and a wish. Real reskilling needs a runway, and IKEA actually budgeted for it.
They kept humans where humans still win. That same Sheffield advisor pointed out that the AI can’t tell which offers a customer has already received. It also misses upsell moments that a person catches. That’s not a dig at Billie. It’s a reminder: the line between AI work and human work is a design choice, not a law of nature.
Here’s what I keep coming back to. Billie was a technology project. What happened to the 8,500 people was a people decision. Somebody had to look at that workforce and see design advisors, not a cost line.
That’s the CHRO’s or CPO’s job now, whether the org chart says so or not. In most companies I talk with, IT and finance still make the AI investment decisions. HR gets invited in afterward to “manage the change.” By then, the cut is already in the model.
The CPOs getting this right bring three things to the table before the tool gets bought:
A clear picture of the work
A credible read on the skills
A potential revenue case for redeployment
At livingHR, we keep calling that the AI-ready CPO, and IKEA is about as clean an illustration as you’ll find.
Partly, and I’ll be honest about that.
Ingka is privately owned, so no analysts ask why headcount didn’t drop after a big AI rollout. That buys patience most public companies don’t have. And IKEA hasn’t been immune to cuts. Both Ingka and Inter IKEA laid off hundreds of corporate workers in spring 2026. Fortune reports that none of those cuts hit the remote sales centers or were tied to AI. Ingka’s chief digital officer, Parag Parekh, hasn’t ruled out future layoffs. But he says they’d more likely stem from economic conditions than from AI.
So no, this isn’t a fairy tale where nobody ever loses a job. It’s IKEA, after all; even the fairy tale comes with an Allen wrench and three mystery screws.
But “we’re not privately held” is just an excuse to move slower, not to skip the work. The approach doesn’t require a Swedish ownership structure. It requires knowing your work, mapping your skills, following the demand, and funding the transition. What it really takes is a people function that brings data to the AI conversation, not just good intentions.
Here’s where I’d start, roughly in the order I walk through with clients.
Build trust, first. Yes, I know trust sounds like a “soft skill.” But McKinsey’s research found that trust in the organization and its leaders is one of the strongest predictors of whether people are ready to use AI. High-trust companies also get more value out of it.
Inventory the work, not the jobs. Break roles into tasks. Figure out which tasks AI will absorb, which it will assist, and which stay stubbornly human. You can’t redeploy what you can’t see.
Study what’s left over. Whatever AI can’t resolve is your demand map. Where are customers (or employees) asking for judgment, expertise, or empathy? That’s where your next roles live.
Build the business case in revenue, not just savings. IKEA’s story lands because €1.25 billion beats €13 million. Bring the growth number to the table, not only the cost number.
Fund the runway. Reskilling that takes two years needs a budget that lasts two years. Treat it like an investment, because it is one.
Tell the story early. Your people will hear “AI” and “efficiency” in the same sentence. If no plan comes with it, they’ll fill in the blank themselves, usually with the worst version.
None of this is glamorous. It’s the unglamorous plumbing behind AI adoption, and it’s exactly why so many AI programs stall after the pilot. The tech works. The people system? Not so much.
The real lesson from IKEA isn’t “be nice to your call center.” The companies getting real value from AI treat it as a workforce design challenge, not a software shopping spree. The chatbot was the easy part. Figuring out what to do with 8,500 people afterward? That was the strategy. But first, you must build trust between your employees and the organization. Read that line one more time.
Every CPO is about to get some version of the IKEA question: “AI just absorbed half this team’s work. Now what?”
The good answer starts with knowing what your people can do next. The bad answer starts with a spreadsheet.
Which one is your organization ready to give?
Billie is an AI customer service assistant that Ingka Group, IKEA’s largest franchisee, launched in 2021. It handles routine questions like store hours, order status, and returns.
No. IKEA retrained about 8,500 call center workers into roles like remote interior design advisor. The corporate layoffs Ingka made in 2026 were not tied to AI, according to Fortune.
IKEA’s remote sales centers brought in €1.25 billion last fiscal year. That compares with roughly €13 million in savings from the chatbot.
Start by building trust. Then inventory the work, not the jobs, and treat whatever AI can’t resolve as a demand map for new roles. Fund reskilling as a multi-year investment.