IKEA Automated 47% of Enquiries. It Retrained 8,500 People Instead of Cutting Them.

The short answer: IKEA's parent company Ingka Group let its AI bot Billie handle about 47% of customer enquiries, roughly 3.2 million interactions, and saved nearly EUR 13 million. It then retrained around 8,500 call-centre co-workers into remote interior design and digital selling instead of cutting them. The savings are the number everybody quotes; the redeployment is the number that actually explains the strategy. Sales through the remote meeting points those advisors staff hit EUR 1.3 billion by the end of FY22. For an SME the headcount is different but the math is the same: automate the repetitive half, decide in advance where the freed hours go, and measure the revenue you gained, not only the cost you removed.

The case study is from 2023, and it has aged better than most things written about AI that year.

Ingka Group, the largest IKEA retailer, deployed an AI customer service bot called Billie. It answers product questions, handles routine enquiries, and runs 24/7 across time zones. Between FY21 and FY23 it resolved around 47% of customer enquiries, about 3.2 million interactions, and generated nearly EUR 13 million in savings.

That is the part that gets screenshotted on LinkedIn.

Here is the part that does not.

Instead of shrinking the call centre, Ingka retrained around 8,500 of those co-workers. They were moved into remote interior design, digital retail selling, relationship building, and complex problem solving. Their global People and Culture manager, Ulrika Biesèrt, framed it as strengthening employability through lifelong learning and reskilling, and accelerating the creation of new jobs.

Two years later, that decision looks less like corporate goodwill and more like the actual business case.

The Number Everybody Quotes Is the Small One

EUR 13 million saved sounds like the headline. It is not.

Sales through IKEA's remote customer meeting points, which is the channel those retrained advisors staff, reached EUR 1.3 billion at the end of FY22. That was 3.3% of total sales, and the company said it wanted to grow that share to 10%.

Read those two numbers next to each other. The AI removed a cost measured in tens of millions. The people the AI freed up went into a channel measured in billions.

That is the whole lesson, and it is the one almost every business gets backwards.

Most AI business cases we see start and end with the cost line. How many hours saved. How many headcount avoided. It is an easy case to write because it is subtraction, and subtraction is easy to model in a spreadsheet.

But subtraction has a ceiling. You can only remove 100% of a cost, and long before you get there the service quality collapses and the savings reverse. Redeployment has no ceiling, because the same hour that used to be spent typing "our showroom opens at 10am" for the four-hundredth time can be spent on a conversation that closes a sale.

IKEA did not automate customer service to have a smaller customer service team. It automated customer service to get 8,500 people into revenue-generating work.

"That Is a Big Company Story. My Business Has Nine Staff."

This is the objection we hear in almost every discovery call, and it is worth taking seriously rather than waving away.

You are right that you cannot run a reskilling programme for 8,500 people. You do not have a learning and development department. You do not have an FY21 to FY23 transformation budget.

You also do not need one, because the ratio is what matters, not the headcount.

Ingka automated roughly half its enquiry volume and moved the freed capacity into selling. If you have three people answering WhatsApp, your version of that is not 8,500 people. It is one and a half people worth of hours, every week, permanently, pointed at work that actually makes money.

For most of the SMEs we work with across Malaysia, Singapore, and Taiwan, that looks like this. The repetitive half is pricing questions, opening hours, "is this in stock", "where are you located", "how does the package work", appointment booking, and the follow-up nobody remembers to send. That is not the valuable part of anyone's job. Nobody joined your company to type it.

The other half is the part where a human is worth paying for: the customer who is comparing you against two competitors, the angry one who is about to churn, the corporate enquiry worth 40 times a normal order, the site visit, the negotiation, the relationship that produces repeat business for a decade.

Small businesses feel this more sharply than IKEA does, not less. When you have three people, losing half of their week to repetitive messages is not an efficiency problem. It is the reason you have not grown.

What Ingka Actually Did, In Order

Strip out the scale and there are three moves. Any business can copy all three.

1. They automated the repetitive share, not the whole job

47% is a very honest number. It is not 100%, and IKEA never claimed it would be.

The AI took the enquiries that are the same every day and left everything else to humans. That is the escalation model, and it is the only pattern we have seen work in real businesses: AI handles the routine, humans handle the exceptions, and the handover carries the full conversation context so the customer does not repeat themselves.

The businesses that fail at this try to automate everything, discover the AI mishandling the 10% of cases that carry 60% of the revenue, and switch it off entirely. We wrote the longer version of this in Will AI Replace My Sales Team?

If you are not sure which half of your workload is the automatable half, our PROVE framework is the five-question test we use with clients to pick the first process.

2. They decided where the people were going before the AI went live

This is the move almost nobody copies, and it is the one that decides whether the project succeeds.

Ingka named the destination in advance: remote interior design and digital selling. The reskilling was planned as part of the deployment, not improvised after the savings showed up.

When the destination is not named in advance, two things happen. The freed hours quietly get reabsorbed into other low-value work within a month, so the automation shows no measurable gain and the renewal conversation gets awkward. And your team, who were never told what happens to them, correctly conclude that the AI is there to replace them, and spend the first two weeks documenting every mistake it makes. That failure pattern is so consistent we can script it before it happens: the AI didn't fail, the rollout did.

Answer the question before go-live, out loud, in front of the team: when the AI takes 40% of your inbox, what are you doing with those hours instead? "Calling back every lead that asked for a quote and never got one" is an answer. "We'll see" is not.

3. They measured the upgrade, not only the saving

Ingka tracked the EUR 13 million saved. It also tracked EUR 1.3 billion in remote sales and set a target to triple that share.

Track both. If your AI reports only tell you how many messages were handled, you have a cost dashboard, not a business case. The questions worth reporting on are: how many leads got a reply within five minutes this month versus last, how many quotes went out, how many lapsed customers were called back, how much revenue came from conversations that would previously have gone unanswered at 11pm.

Cost saved is what pays for the system. Revenue added is what makes you keep it.

The Honest Part: This Took Three Years

Billie's rollout started in FY21. The 47% and the EUR 13 million are cumulative through FY23.

That is not a story about a bot switched on in a weekend. It is three years of an AI system being corrected, extended, and taught the business, alongside a workforce being retrained around it.

An SME moves faster than IKEA because there is less to coordinate, but the shape is the same. In week one the AI will get things wrong, because it does not yet know your business the way your best employee does. That knowledge only enters the system through testing and correction, which is the real bottleneck in the AI age. Businesses that expect week-one perfection quit in week two. Businesses that treat the AI Employee like a new hire, coached and reviewed and improved, compound.

What This Looks Like In Your Business

Take the IKEA structure and shrink it to your size.

The repetitive layer goes to an AI Employee. Instant replies on WhatsApp, Instagram, Messenger, LINE, and Telegram, at midnight and on Sunday. Qualification questions asked consistently. Appointments booked. Follow-up sent on schedule instead of when someone remembers.

The human layer gets upgraded. Your staff stop being a reply queue and start being closers, advisors, and relationship owners. The AI hands them the conversation with the context already loaded, so they arrive knowing what the customer asked and what was promised.

The management layer gets visibility. The reason IKEA could report those numbers is that the system produced them. The same applies here: the AI Manager compiles what is happening across every conversation, so you know which objections keep coming up and which follow-ups are leaking, instead of guessing.

That is what "AI transformation" actually means. Not a bot on the website. A redesign of who does what, with the repetitive work moved to a system that never gets tired and the human work moved up the value chain.

Key takeaways:

1. The savings are the small number. IKEA saved nearly EUR 13 million with AI and staffed a EUR 1.3 billion remote sales channel with the people it freed up.
2. 47% is the realistic target, not 100%. AI handles the routine share, humans handle the exceptions, and the handover carries full context.
3. Name the destination before go-live. If nobody has decided where the freed hours go, they disappear into low-value work and your team assumes the worst.
4. Measure the revenue added, not only the cost removed. Cost saved pays for the system; revenue added is why you keep it.
5. The ratio scales down. Your version of 8,500 people is three people, and the mechanics are identical.

The businesses that will struggle with AI are not the ones that adopt it late. They are the ones that adopt it as a cost-cutting exercise, get their EUR 13 million equivalent, and stop there.

Frequently Asked Questions

What is IKEA's Billie AI chatbot?

Billie is an AI customer service bot deployed by Ingka Group, the largest IKEA retailer, starting in FY21. It uses natural language processing to answer product questions and handle routine enquiries 24/7 across time zones. Between FY21 and FY23 it resolved around 47% of customer enquiries, about 3.2 million interactions, and generated nearly EUR 13 million in savings.

Did IKEA cut jobs because of AI?

No. Ingka Group retrained around 8,500 call-centre co-workers rather than reducing the team. They were reskilled into remote interior design, digital retail selling, relationship building, and complex problem solving, which are the parts of the job the AI could not do. The company framed it as strengthening employability through lifelong learning and accelerating the creation of new jobs.

How much did IKEA save with its AI chatbot?

Nearly EUR 13 million across FY21 to FY23, from roughly 3.2 million AI-handled interactions. The larger business number was on the revenue side: sales through remote customer meeting points, the channel the retrained advisors staff, reached EUR 1.3 billion by the end of FY22, or 3.3% of total sales, with a stated ambition of 10%.

What should a small business do with the staff time that AI frees up?

Decide before go-live, not after. Point the freed hours at work that only humans can do and that produces revenue: calling back high-value leads, quoting, site visits, negotiation, retention calls to lapsed customers. If nobody has named the destination for those hours, they get reabsorbed by low-value work within a month and the automation shows no measurable gain.

Does AI reskilling work for a small business with only a few staff?

Yes, because the ratio matters more than the headcount. IKEA moved 8,500 people; an SME with a three-person team is moving three people. The mechanics are identical: automate the repetitive enquiries first, redeploy the freed hours into revenue work, and measure the revenue lift rather than only the cost saving.

Source: Ingka Group newsroom, "AI and remote selling bring IKEA design expertise to the many", published 29 June 2023.

Want to know which half of your team's inbox an AI Employee could take over? Book a free strategy call and we will map the repetitive work, the escalation points, and where the freed hours should go, before anything gets deployed. We have done this for 600+ businesses across Malaysia, Singapore, and Taiwan.