Ikea automated customer service and kept the people

Ikea's AI chatbot Billie now assists 74% of customers, up from 47%, and the company retrained about 8,500 call center staff as interior design advisors, leading to customer satisfaction rising from 60% to 89% and remote sales reaching €1.25 billion in 2025, up from €1.08 billion. The World Economic Forum estimates 92 million jobs could be displaced by AI by 2030, but Ikea's example shows that automation can improve service and create value when freed workers are redeployed to higher-value roles.

Ikea automated customer service and kept the people In 2021 Ikea put a chatbot on customer service and named it Billie, after the bookcase. It handled the questions you would expect: opening hours, delivery windows, whether you can bring a dog into the store. Fortune reported https://fortune.com/2026/07/30/ikea-ai-workforce-reskilling-jobs-billie-chatbot-global-500/ that Billie now assists 74% of customers, up from 47%. That’s the point in the story where you brace for the layoffs. They didn’t happen. Ikea retrained roughly 8,500 call center staff as interior design advisors, in programs lasting five or six weeks, and pointed them at remote sales. Those remote sales centers, 24 of them supporting 31 countries, brought in €1.25 billion last year, up from €1.08 billion, growing between 15 and 20% a year. And customer satisfaction went from 60% to 89%. The number that changes the story The satisfaction figure is the surprising one, because it runs counter to expectations. Automating service normally means accepting worse service in exchange for lower cost. You wait longer, you talk to something that misunderstands you, and the company books the saving. Ikea’s went the other way, and the mechanism isn’t mysterious. The people who used to answer “what time do you close” now help you plan a kitchen. That’s a better conversation for the customer and a considerably more valuable one for Ikea. The chatbot didn’t directly improve service. It moved people off work that didn’t need them, which improved service in a uniquely human way. Two caveats. Assisting 74% of customers is not the same as resolving 74% of their problems, and the gap between those could be wide. And these are Ikea’s own numbers, reported by them, about a program they are proud of. Trust the direction more than the exact figures. Automation only pays if you have somewhere to put the people The lesson here’s not that AI creates jobs, nor is it that AI destroys them. Both of those arguments are conducted at a level of abstraction where nothing can be decided. The lesson is narrower and more useful: automating routine work only pays if you have somewhere better to put the capacity it frees. Most companies treat automation as a cost line. Work goes away, headcount goes away, the saving lands in the budget, and that’s the end of the story. The saving is there, and it’s also the smallest possible version of the opportunity, because the ceiling on cost-cutting is the cost itself. You can only ever save what you were spending. Ikea treated it as a capacity question instead. The same technology, the same freed hours, and a different decision about where they went. One version produces a smaller call center. The other produces a €1.25 billion sales channel that grows 15 to 20% a year. The World Economic Forum’s estimate, cited in the same piece, is that 92 million jobs could be displaced by AI by 2030. Whether that number lands as a catastrophe or a reallocation depends almost entirely on how many companies make the decision Ikea made, and how few make the other one. That other choice has a long history /blog/most-tech-revolutions-made-work-worse-for-employees : every productivity wave since the PC made work faster and then filled the time back up, so the companies gained more than the people did. What this looks like for a 40 person company Ikea had 8,500 people to redeploy and a ready destination for them. Almost nobody reading this does. At a smaller company the same decision arrives in a different form. Nobody’s job title is “read the inbox”. Nobody is employed to work out which of yesterday’s messages contained a commitment. So when that work goes away, you are not redeploying anyone and no role disappears. What you get back is hours, spread thinly across people who were already doing three jobs. Which makes the question harder rather than easier, because the freed hours have no obvious destination. They get absorbed. The week fills up the way it always did, and six months later nobody can point at what changed. The work those hours should go to is usually the work that was never getting done at all, rather than work that was being done by someone else. The follow-up nobody made. The customer who churned quietly because no one called. The proposal that went out two days late every time. Every small company has a list of things everyone agrees matter and nobody has time for, and that list is the actual answer to what automation is for. Decide before, not after The practical version of all this is a question worth answering in advance. If something gave your team back six hours a week each, what would you want those hours spent on? Not in principle. Specifically, this quarter, with names attached. Most leaders have not decided, which means the hours get absorbed by whatever is loudest. That is the difference between Ikea’s outcome and the ordinary one, and it is a decision about the destination rather than a decision about the technology. this+that’s role in automation We build software that does routine work for people, so we have a stake in the argument. this+that reads the messages arriving in the channels a team already uses, pulls out the requests and commitments buried in them, and runs the workflows the team described in plain language. The effect is hours returned, mostly in small pieces: the triage you didn’t do, the follow-up you didn’t have to remember, the reply you didn’t have to write from scratch. We are not going to tell you that this creates a €1.25 billion sales channel. What we would say is that those hours go a lot further when you have decided in advance where they should go. That decision is yours, it costs nothing, and in our experience it is what decides whether the gain compounds or the team just ends up feeling like it is getting vaguely more done. Key takeaways - Ikea’s chatbot assists 74% of customers, up from 47%, and customer satisfaction rose from 60% to 89% rather than falling. - The gain came from moving roughly 8,500 call center staff into interior design sales, not from the automation itself. Those remote sales centers now turn over €1.25 billion, growing 15 to 20% a year. - Automating routine work only pays if you have somewhere better to put the freed capacity. Cost-cutting caps your upside at what you were already spending. - At a small company no one is redeployed, because nobody’s job was the routine work. You get back hours, and hours are absorbed unless they have a named destination. - The hours should go to the work that was never getting done, rather than to work someone else was doing. - Decide where the time goes before you free it up. Most companies decide afterwards, which means they don’t decide.