AI for Small Business: What IKEA Actually Did

There is a graphic going around. Blue IKEA store, bold yellow type, three numbers stacked up: 8,500 roles automated, 0 people laid off, $1.4 billion in new revenue. Underneath, the question everybody wants answered: what if AI’s biggest opportunity is not replacing people but repositioning them?

The short version: the underlying story is real, well documented, and genuinely one of the better examples of AI for small business owners to learn from. But of the three numbers on that card, one is right, one is right only if you read it narrowly, and one is wrong in a way that matters. We checked all three against IKEA’s own newsroom and current reporting. Here is what actually happened, and the specific move worth copying.

What the graphic gets wrong

Start with the corrections, because the corrected version is the more useful story.

“8,500 roles automated” is backwards. Ingka Group, the largest IKEA franchisee, states in its own announcement that “8,500 call centre co-workers have been reskilled.” Nobody’s role was automated out of existence. The chatbot, called Billie after the Billy bookcase, automated a share of incoming contacts, not a headcount. That share was about 47% of enquiries between 2021 and 2023, and Fortune reports it now sits near 74%. Collapsing “8,500 people retrained” into “8,500 roles automated” inverts the whole point.

“0 people laid off” is true of this program and misleading about the company. Fortune is explicit that none of the layoffs touched the remote-sales centers and none were attributed to AI. So the reskilling effort itself really did cut nobody, which is the impressive part. But IKEA did make cuts in 2026: roughly 800 office roles at Ingka in March and about 850 at Inter IKEA in May, around 1,650 in total. Those were corporate and group function roles, blamed on consumer confidence, tariffs and organizational complexity rather than automation. Different workforce, different cause. Worth knowing before you repeat the number.

“$1.4B in new revenue” is the real problem. The dollar figure is a fair currency conversion. Ingka reported remote selling at EUR 1.3 billion at the end of FY22, or 3.3% of total sales, and Fortune’s 2026 reporting puts the most recent fiscal year at EUR 1.25 billion, up from EUR 1.08 billion. Figures wobble a little between reporting years and sources. But none of that is new revenue created by a chatbot. It is the total revenue of an entire sales channel that already existed and was already growing. The year over year increase is closer to EUR 170 million. Handing the full amount to the bot is a causal leap, and it is the claim most likely to get you challenged if you repeat it in a meeting.

Directionally true, numerically loose. The corrected story is still remarkable.

So what actually happened?

Strip the exaggeration and the sequence is clean.

IKEA launched Billie in 2021 to handle the bottom tier of customer contact: opening hours, order status, stock checks, return policy. The dull, repeating questions. Ingka says that between 2021 and 2023 the bot resolved about 3.2 million interactions and saved close to EUR 13 million.

Note the size of that saving. Thirteen million euros, against a company turning over tens of billions. As a cost-cutting exercise it is a rounding error. If saving money had been the goal, this would be a forgettable project.

Instead, IKEA took the 8,500 people whose day had just been freed up and retrained them as remote interior design advisors. Fortune reports the training runs five to six weeks. Those advisors now sell: they plan rooms, they handle complex problems, they build relationships with customers who are spending real money on kitchens and storage. The remote channel served around 10 million customers last fiscal year. Ingka’s chief digital officer, Parag Parekh, described the growth curve as a hockey stick. Customer satisfaction went from 60% before Billie to 89%.

The target is for remote selling to reach 10% of total revenue. It was 3.3%.

Why it worked, and it is not the chatbot

Here is the part the viral card misses entirely.

The call center was a cost center staffed by the only people in the building who already knew the product catalog cold. They spent their days telling customers what time the store closed. That is an extraordinary waste of accumulated knowledge, and it was invisible because it looked like a functioning department.

Billie did not create EUR 1.25 billion. Billie created availability. It freed 8,500 people who already had the expensive, slow-to-acquire part of the skill set, product knowledge, and left them needing only the fast part, sales and design training, which took about six weeks.

That is the transferable insight, and it is worth sitting with. In most businesses the constraint is not that nobody can do the high-value work. It is that the people who could do it are buried in work that repeats.

The second thing IKEA got right is sequencing. They had somewhere to put the freed capacity. The remote design channel existed before the bot did. Automate without a destination for the released hours and you do not get a hockey stick, you get idle time, and idle time is what turns into a layoff six months later. The destination is the strategy. The bot is just the tool that opens the door to it.

The playbook at small-business scale

You do not need 8,500 people for this to apply. You need two or three, and one repeating job. Here is the same move, sized down.

1. Count the bottom tier for one week. Tally every inbound question that repeats and requires no judgment. Hours, availability, pricing, “did my order ship,” “where are you located,” “do you take card.” Do not estimate it, count it. Most owners are shocked by the number, and the count is what tells you whether automation is worth anything here at all.

2. Name the destination before you automate anything. Write down, in one sentence, what the freed hours will be spent on and what that produces. “Sarah gets six hours a week back to follow up on quotes that are sitting unanswered” is a destination. “We will be more efficient” is not. If you cannot write the sentence, you are not ready to automate, because you have nowhere to put what you release. This is the step almost everyone skips.

3. Check who already holds the expensive knowledge. IKEA’s advantage was that its call center staff already knew the products. Six weeks of training completed them. Look at who in your business already knows the customers, the pricing quirks, the reasons jobs go wrong. That person is your design advisor. They are probably answering the phone. If you want a structured way to see which of your work is routine enough to hand off and where your people should be pointed instead, our AI Readiness Assessment walks through it in a few minutes.

4. Automate the tier, never the person. Point the tool at the category of question, and keep a clean handoff to a human the moment it stops being routine. Our guide to AI customer service for small business covers the scope and handoff rules that decide whether this works or embarrasses you.

5. Measure the destination, not the savings. IKEA’s EUR 13 million saving was trivial. The channel it enabled is worth roughly a hundred times that. If you only track hours saved, you will conclude the project was marginal and quietly stop. Track what the freed capacity produced: quotes followed up, jobs closed, repeat customers. That is the number that justifies going further.

Where the comparison breaks

Three honest limits, because a playbook that only lists upside is a sales pitch.

IKEA is enormous and had years, plus an existing remote channel to grow into. Most small businesses are building the destination and the automation at the same time, which is harder and slower.

The causal link is loose. Remote selling was growing before Billie and would likely have grown without it. The bot accelerated a trend rather than starting one. Be suspicious of anyone, including us, drawing a straight line from a tool to a revenue figure.

And the 2026 layoffs are a genuine asterisk. They tell you that a company can do the reskilling thing sincerely in one part of the business and still cut roles elsewhere when trading conditions turn. Reskilling is a strategy, not a promise, and it is worth being clear-eyed about that rather than selling it as a guarantee.

None of that undoes the lesson. It sharpens it. The reason to reposition people rather than remove them is not sentiment, it is that the people who know your customers are the hardest thing in your business to replace and the easiest thing to waste. IKEA noticed that its most product-literate staff were spending their days reciting opening hours, and it did something about it. That is the whole play. If you want a starting sequence for the first month, our 30-day path for using AI in a small business is the practical companion to this one, and our piece on outsourcing covers how to sort which work can be handed off in the first place.

Frequently Asked Questions

Is the viral IKEA statistic accurate?

Partly. The 8,500 figure is real but describes people who were reskilled, not roles that were automated away. No one in that group was laid off, and none of IKEA’s 2026 layoffs were attributed to AI or affected the remote-sales centers. The revenue claim is the weakest link, because the roughly 1.25 billion euro figure is the total revenue of the entire remote selling channel rather than new revenue produced by the chatbot.

What did IKEA’s Billie chatbot actually do?

Billie launched in 2021 to handle routine, high-volume customer questions such as store opening hours, order status, stock availability and return policy. Ingka reports it resolved about 47% of enquiries between 2021 and 2023, roughly 3.2 million interactions, saving close to 13 million euros. Fortune reports that share has since risen to around 74%. It was pointed at a category of question, not at a group of employees.

Why did IKEA retrain workers instead of cutting costs?

Because the saving was small and the opportunity was large. The 13 million euros Billie saved is negligible for a company of IKEA’s size, while the call center staff represented thousands of people who already knew the product catalog in depth. Retraining them as remote interior design advisors took five to six weeks and turned a cost center into a sales channel that now serves around 10 million customers a year.

Can a small business copy this approach?

Yes, at a much smaller scale, and the sequence matters more than the size. Count the repeating questions that require no judgment, decide in advance what the freed hours will be used for, identify who already holds the hard-won knowledge about your customers and products, automate the category of task rather than the person, then measure what the freed capacity produced rather than only the hours saved. The step most businesses skip is naming the destination before automating.

We keep coming back to one question here, and we would genuinely like to hear your answer: who in your business already knows the most about your customers, and how much of their week is currently spent on questions that repeat?