Autumn 2021. Somewhere inside an IKEA customer contact centre, an agent hangs up after explaining, for the thirtieth time that day, where a bookcase delivery stands. A new colleague has just appeared on the screen. His name is Billie. He never sleeps, never takes a break, and will soon absorb almost one customer request out of two. In most companies, that scene is the opening act of a layoff plan. At IKEA, it came right before the birth of a billion-euro revenue channel.
I often say that panic in the face of AI is directly linked to the survival instinct. When an executive looks at a tool like Billie, the brain does not see an opportunity. It sees a threat to 8,500 jobs. IKEA looked at the same screen and asked a different question. That gap is the whole difference between a company that agonizes and a company that innovates.
The reflex nine executives out of ten would have had
Let us run the exercise together. You run customer service. You learn that a conversational assistant will handle 47 % of requests, roughly 3.2 million interactions a year (Ingka). You employ 8,500 people whose work largely consists of answering exactly those requests. On a scale from 0 to 5, how strong is the pull to cut headcount? Be honest. The “how many jobs can I remove” reflex is a short-term financial survival reflex.
That reflex rests on a bias I hunt in every talk I give: confusing the task with the person. The repetitive task, the order tracking or the product return, can be automated. The person who used to perform it does not boil down to that task. Executives who forget this are quietly setting up a disaster. A company that fails to innovate in how it treats its people is already dead, it just does not know it yet.
The question IKEA asked instead
IKEA started by mapping what AI could not do: understanding a fuzzy need, reassuring an anxious customer, imagining how to fit a family of four into 45 square metres. It then analysed the calls and found something Billie could not capture. Many customers were not calling to get an answer. They were calling to be guided (CIO).
From there, the decision becomes almost obvious. IKEA trained its 8,500 call-centre workers as remote interior design advisors, across 24 centres in 31 countries (Fortune). Five to six weeks of training for new joiners. Billie takes the repetitive load, the human moves up a level toward advice, remote selling and relationship. Customer satisfaction rose from about 60 % to 89 % (Fortune). You do not manufacture that number with a chatbot. You manufacture it with humans you have freed from boredom.
The real maths: 13 million saved, 1.3 billion created
Here is my favourite figure in the whole story, because it dismantles the obsession with cost. Billie, the tool, generated roughly 13 million euros in savings between 2021 and 2023 (CIO). Interesting, yet modest at IKEA’s scale. The remote-selling channel run by the 8,500 reskilled advisors reached 1.3 billion euros in sales by the end of fiscal 2022, about 3.3 % of revenue, with a target of 10 % in the following years (Ingka). On the latest known year, that channel is worth around 1.25 billion euros, close to 1.4 billion dollars, growing 15 to 20 % a year (Fortune).
Look at the orders of magnitude. The tool saved millions. The humans, placed at the right spot, created a billion. A factor of one hundred separates them. When I say stop asking “how do I replace five people with AI” and start asking “how can five people do the work of fifty with AI”, this is what the answer looks like in numbers. I unpack that shift and its method in (My book, chapter 14).
Where the story gets complicated, and that is a good thing
I would be a poor speaker if I stopped at that beautiful curve. My first instinct in front of a magnificent result is always the same: is it realistic, is it verifiable? So let us verify.
In 2026, IKEA did carry out layoffs. Ingka cut about 800 office roles in March, Inter IKEA about 850 in May, close to 1,650 positions (EmployerBranding.news). Two nuances matter. These cuts hit corporate functions, not the 8,500 reskilled advisors. And they were attributed to declining sales and tariffs, not to AI. Ingka’s chief digital officer, Parag Parekh, did not rule out future adjustments driven by the economy (PYMNTS).
What should you take from this? That reskilling is not an eternal life insurance against layoffs. It was a decision, made at a moment, with clear values. A remarkable decision does not vaccinate a company against a tough market. I would rather hand you that honest version than a fairy tale. People trust you more when you point out the grey zones in what you share yourself.
Replace tasks, amplify people
Let us return to the principle, because one isolated case is not a strategy. When an activity is repetitive, the first question is never “who will do it”. It is “can we remove it, automate it, or hand it to a machine”. A well-designed process does not chase humans out of the company. It moves them toward what they do best: thinking, creating, deciding, solving the unexpected.
This conviction has a name in my system: the invariant of innovation is human, never technological. Technology changes constantly, its only invariant being the acceleration of its own change. What stays constant are the fears, the biases and the emotions of people facing change. António Damásio put it better than I can: we are feeling machines that think before we are thinking machines that feel. IKEA treated the Billie affair as a human problem before treating it as a technical one. That is the opposite of what the digital toy collectors do (Stanford HAI). The full approach, with its five pillars, is laid out in (My book).
The method: measure before you believe
You will tell me that everyone promises AI gains. Precisely. The famous BCG–Harvard study on 758 consultants is quoted everywhere, 25.1 % faster across 18 tasks. The trap is believing that number applies to every function. Another study covering about 25,000 workers across 7,000 workplaces brings everyone back to earth: the average time actually saved sits around 2.8 to 3 %. The gap between 25 % promised and 3 % observed is where disillusion is born, and sometimes a strike. Assumptions are dangerous, so kill them with experimentation as early as possible.
In practice, IKEA followed, without knowing it, the logic I teach in five steps: a clear vision of what AI changes, honest communication to the teams with a baseline sentiment measurement, an exploration team to test, a measured rollout, then a sentiment measurement at the finish. Peter Drucker warned that culture eats strategy for breakfast. The Billie strategy worked because the culture followed, with real training resources behind it. Ask yourself, on a scale from 0 to 5: in your organization, do you truly measure a gain before, during and after, or do you simply believe in it?
The essentials
Remember three things:
- AI replaces tasks, not whole professions. The value lever is placing the freed-up humans where they create the most, the way IKEA turned 8,500 agents into advisors.
- The tool saves, the humans create. Billie saved 13 million, the reskilled advisors generated over a billion. A factor of one hundred should decide your investment priorities.
- A brave decision is not a permanent guarantee. IKEA also cut jobs in 2026 for market reasons. Measure, stay honest, and never buy a productivity promise without proof.
Innovating in the face of AI is not a technological sprint. It is teamwork that starts with a vision and is verified by measurement. Innovate or agonize, the choice is up to you.
References
- (My book) https://philippeboulanger.com/book/
- (Ingka) https://www.ingka.com/newsroom/ai-and-remote-selling-bring-ikea-design-expertise-to-the-many/
- (CIO) https://www.cio.com/article/4180896/how-ikea-turned-a-chatbot-with-13-million-users-into-a-business-worth-1-3-million.html
- (Fortune) https://fortune.com/2026/07/30/ikea-ai-workforce-reskilling-jobs-billie-chatbot-global-500/
- (PYMNTS) https://www.pymnts.com/news/artificial-intelligence/2026/ikea-turned-8500-call-agents-into-design-consultants/
- (EmployerBranding.news) https://employerbranding.news/resources/what-ikeas-ai-reskilling-story-looks-like-now-it-has-made-layoffs-too/
- (Stanford HAI) https://aiindex.stanford.edu/report/
