The $1.4 Billion Difference

How IKEA Chose Augmentation Over Replacement

Twin Ladder Casebook | Twin Ladder | February 2026

The Hook

Maria had worked in IKEA customer service for six years. She knew every return policy, every delivery timeline, every answer to "Where is my order?" She could recite the KALLAX shelf dimensions from memory. She was, by every metric the company tracked, excellent at her job.

Then, in 2021, a chatbot named Billie arrived and started answering those same questions --- thousands of them, simultaneously, in dozens of languages, without coffee breaks or sick days. Maria watched the system handle 47 percent of all customer inquiries in its first operating period. The math was not subtle. If a machine could do nearly half her job today, the trajectory was clear.

Except IKEA did not follow that trajectory. Instead of handing Maria a severance package, the company handed her a training program. Within months, she was no longer answering "Where is my order?" She was answering "How do I make my 42-square-metre apartment feel like a home?" She had become an interior design advisor --- and IKEA had found $1.4 billion in revenue it did not know it was missing.

This is the story of what happens when a company asks the right question about AI. Not "How many people can we replace?" but "What can our people do when the machine handles the routine?"

The Story

The decision did not emerge from a single boardroom meeting. It unfolded over three years, beginning in 2021, when Ingka Group --- the holding company that operates the majority of IKEA stores worldwide --- deployed Billie, a natural language processing chatbot built to handle frontline customer service inquiries. The rationale was straightforward: a significant portion of customer contacts involved repetitive, transactional questions. Delivery status. Return procedures. Store hours. Product availability. These were questions with definitive answers, questions that did not benefit from human nuance, empathy, or creativity.

Billie performed. Between 2021 and 2023, the chatbot resolved approximately 3.2 million customer interactions, handling 47 percent of all inquiries it received. The savings were tangible --- nearly 13 million euros in operational cost reduction, according to Ingka Group's own reporting. By any conventional efficiency metric, the deployment was a success.

Here is where the story diverges from the standard automation narrative. Most organizations, having demonstrated that a machine can do a task cheaper and faster, proceed to reduce headcount. The logic feels irresistible: if the chatbot handles half the calls, you need half the people. Ingka Group did not follow that logic. Instead, CEO Jesper Brodin and the leadership team made a decision that would prove to be worth billions: they would retrain the workers whose tasks Billie had absorbed.

The reskilling program began in 2021 and eventually encompassed 8,500 call centre co-workers across multiple markets. These were not token workshops. The training covered remote interior design competence, digital retail sales techniques, relationship-building skills, and the capacity to handle complex customer inquiries requiring creative problem-solving. Workers who had spent years answering "Is the MALM dresser in stock?" were trained to conduct one-on-one virtual consultations --- full design sessions conducted via phone or video, where they would develop mood boards, floor plans, 3D renderings, lighting plans, and curated product lists for individual customers.

The commercial results were extraordinary. In Ingka's fiscal year 2022, sales generated through the remote interior design channel --- the channel staffed by reskilled former call centre workers --- amounted to 1.3 billion euros, approximately $1.4 billion. That figure represented 3.3 percent of Ingka's total revenue, a revenue stream that had been negligible before the reskilling initiative. Ingka set a target to grow this share to 10 percent by 2028.

Consider what happened here. IKEA did not merely avoid the cost of layoffs. It did not merely preserve goodwill or burnish its employer brand. It discovered an entirely new revenue stream --- $1.4 billion --- by redirecting human capability toward work that AI could not do. The chatbot answered the transactional questions. The humans answered the questions that required taste, spatial reasoning, emotional sensitivity, and the ability to understand what a customer meant when they said, "I want my living room to feel calm."

The economics deserve scrutiny, because they reveal something counterintuitive. The conventional wisdom holds that AI adoption is a cost story: deploy the technology, reduce headcount, improve margins. IKEA turned it into a revenue story. The 13 million euros saved by Billie's automation were real, but they were rounding errors compared to the 1.3 billion euros generated by the reskilled workforce. The cost savings represented the floor --- what any company with a chatbot could achieve. The revenue generation represented the ceiling --- what only a company willing to invest in its people could reach. The ratio is roughly 100 to 1. For every euro saved by replacing human effort with AI, IKEA earned one hundred euros by redirecting that human effort toward higher-value work.

Brodin himself has described generative AI as "the biggest opportunity" and "the iPhone moment for enterprise," while simultaneously acknowledging its dangers. He has noted that "most of the jobs they are doing right now are soul crushing" --- a remarkably candid admission from a CEO that the work being automated was not work that developed human capability in the first place. The reskilling was not charity. It was strategic recognition that the humans were more valuable doing different work than they were doing no work at all.

There is a subtlety here that most analyses of the IKEA case overlook. The 8,500 workers did not become interior design advisors despite their call centre experience. They became effective advisors because of it. Years of fielding customer questions had given them an intimate understanding of how IKEA customers think, what they struggle with, what language they use to describe their homes, and what frustrations drive them to call in the first place. That accumulated knowledge --- tacit, unstructured, impossible to encode in a training dataset --- became the raw material of the design advisory service. IKEA did not waste six years of customer insight. It monetised it.

The ambition extends beyond the initial 8,500. IKEA announced plans to train 70,000 employees in AI literacy by 2026 --- a commitment to building foundational competence across the entire workforce, not merely within the group directly affected by chatbot deployment. The interior design service expanded from its initial European and Australian markets to the United States and the United Kingdom, with consumer design consultations priced at $99 per room. The service is not a loss leader. It is a profit centre staffed by people who, three years earlier, were reading delivery tracking numbers aloud.

Through the Twin Ladder Lens

The IKEA case is a near-perfect illustration of the Twin Ladder in action --- specifically, the transition from Level 0 to Level 1.

In the Twin Ladder framework, Level 0 represents the AI Literacy Foundation: the baseline ability to critically evaluate AI output, to understand what AI can and cannot do, and to distinguish between tasks suited for automation and tasks that require human judgment. IKEA's leadership demonstrated Level 0 competence when it assessed Billie's capabilities honestly. The chatbot could resolve 47 percent of inquiries. That meant it could handle the transactional, repetitive, pattern-matched work. It also meant --- and this is the critical insight --- that it could not handle the remaining 53 percent, the interactions requiring contextual understanding, creative thinking, and emotional intelligence.

Level 1, the Professional Twin, is where the organization mirrors individual roles with AI agents --- not to replace professionals but to create a productive division between human and machine. The purpose is comparison and liberation: the AI handles what it does well, and the human is freed to operate at a higher cognitive level. The emotional arc is predictable: fear (the chatbot is doing my job), curiosity (what else could I do?), relief (my judgment still matters), and growth (I am doing more valuable work than before).

IKEA's 8,500 reskilled workers followed exactly this arc. They began with a legitimate fear that Billie would make them redundant. They moved through a structured training program that redirected their deep product knowledge and customer empathy toward design advisory work. They arrived at a new professional identity --- interior design advisor --- that was more complex, more creative, and more commercially valuable than the role AI had absorbed.

The key architectural principle of Level 1 is that the twin must preserve domain competence, not erode it. IKEA achieved this by ensuring that the reskilled workers did not simply defer to AI tools in their new roles. They brought years of accumulated customer insight --- knowledge of what customers actually buy, what frustrates them, what delights them --- into the design advisory function. That tacit knowledge, built through thousands of customer interactions, became the foundation of a service that no chatbot could replicate. The workers did not lose competence. They transferred it upward.

This is also where the Competence Paradox, as documented in Twin Ladder's research, becomes visible in its positive form. The Competence Paradox warns that AI tools which accelerate individual performance can simultaneously degrade the human capabilities organizations depend on. IKEA avoided this trap by design. Rather than allowing workers to become passive operators of an AI system --- the pattern that produced the 21 percent decline in adenoma detection among gastroenterologists documented in The Lancet --- IKEA moved workers into roles where active human engagement was the product. The chatbot handles "Where is my order?" so the human can handle "Help me design my living room." One is a retrieval task. The other is a judgment task. The distinction is everything.

There is a broader lesson embedded in the framework here. The Twin Ladder describes four levels --- from Level 0 (AI Literacy) through Level 3 (Ecosystem Twin). Most organisations fixate on the upper levels, attempting to build sophisticated digital twins of their operations and supply chains. IKEA's genius was its discipline in executing Level 0 and Level 1 thoroughly before reaching higher. The company's plan to train 70,000 employees in AI literacy by 2026 is a Level 0 investment. The reskilling of 8,500 workers into design advisors is a Level 1 investment. Neither is glamorous. Neither generates headlines about cutting-edge technology. Both generated billions in value. The ladder is climbed, not skipped --- and IKEA proved why.

The Pattern

Three hundred kilometres north of IKEA's Almhult headquarters, in Stockholm, another Swedish company made a different choice at almost exactly the same time --- and the contrast is instructive.

In February 2024, Klarna announced that its AI assistant, built in partnership with OpenAI, had taken over 75 percent of customer service chats, handling approximately 2.3 million conversations. The company declared that the system was doing the work of 700 employees. Between 2022 and 2024, Klarna's headcount fell from 5,527 to approximately 3,400 --- a reduction of nearly 40 percent. CEO Sebastian Siemiatkowski framed this as progress: AI was making the company leaner, faster, more efficient.

The market initially applauded. Then the customer feedback arrived.

By early 2025, internal reviews revealed what the efficiency metrics had obscured. Customers reported generic, repetitive, and insufficiently nuanced responses. Complaint volumes increased. User satisfaction ratings declined. The AI could process transactions, but it could not process frustration, ambiguity, or the particular tone of a customer who had been incorrectly charged and needed someone to understand why that mattered. Siemiatkowski publicly acknowledged that Klarna had "gone too far," admitting that the focus on efficiency had eroded the quality of the company's service. "Obviously, AI today can pretend to be empathetic and express emotions," he told Bloomberg, "but at the core, people crave human connection."

Klarna reversed course. The company began rehiring human customer service agents, specifically targeting students and rural workers for on-demand remote roles. The reversal came days after Klarna's high-profile US IPO, which valued the company at $19.65 billion. The timing was not coincidental --- investors and customers had both signalled that pure replacement was not a sustainable strategy.

Two Swedish companies. Same era. Same technological moment. Opposite strategies. Opposite outcomes.

IKEA asked: "What can our people do that the machine cannot?" It found $1.4 billion in new revenue. Klarna asked: "How many people can the machine replace?" It found a customer satisfaction crisis and a public reversal.

The difference was not technological. Both companies had access to the same AI capabilities. The difference was philosophical. IKEA treated AI as a tool for human elevation. Klarna treated AI as a substitute for human presence. The market punished one approach and rewarded the other --- not because the market is sentimental about employment, but because it correctly assessed which approach produced better outcomes.

The Lesson

The IKEA-Klarna comparison is not an isolated anecdote. It is a pattern that will define the next decade of organizational performance. The question is not whether AI can perform a given task. In an increasing number of domains, it can. The question is what happens to organizational capability when you remove the humans from the equation --- and what happens when you redirect them instead.

Augmentation outperforms replacement. This is not an ideological claim. It is an empirical observation supported by IKEA's $1.4 billion in new design advisory revenue, by Klarna's forced reversal, and by the broader evidence documented in the Competence Paradox research: the Lancet study showing degraded medical performance after AI dependency, the NASA research documenting pilot skill erosion, the Microsoft data revealing collapsed cross-functional collaboration in AI-heavy workflows.

The economics are unambiguous. IKEA's approach created value. Klarna's approach destroyed it, then required additional investment to reverse. The total cost of Klarna's replacement strategy --- the savings from headcount reduction minus the cost of customer attrition, reputation damage, and subsequent rehiring --- has not been publicly disclosed, but the trajectory is clear. Cutting costs by eliminating people is arithmetic. Finding new revenue by elevating people is strategy. Arithmetic is easy to execute and easy to copy. Strategy is neither.

The organizations that will lead are not the ones that automate fastest. They are the ones that answer a harder question: "When AI handles the routine, what do we train our people to do next?" IKEA answered that question. It looked at 8,500 call centre workers and saw not a cost to be eliminated but a capability to be redirected. It invested in training, built a new service category, and generated a revenue stream that did not exist before.

Every organisation facing AI deployment confronts the same fork in the road. One path leads to headcount reduction, short-term margin improvement, and the gradual erosion of the human capabilities that differentiate the company from its competitors. The other path leads to reskilling, new service categories, and revenue streams that did not previously exist. The first path is easier to model in a spreadsheet. The second path is harder to execute but vastly more rewarding. IKEA proved this at scale, across multiple markets, with measurable results.

The Twin Ladder framework provides the structure for this thinking. Level 0 establishes the literacy to know what AI can and cannot do. Level 1 creates the partnership between human and machine that frees human capability for higher-order work. IKEA climbed. Klarna skipped. The $1.4 billion difference is the distance between those two choices.

The Monday Morning Question: Look at the five most common tasks your team performs this week. Which of those tasks could a well-trained AI handle --- and what would your people do with the hours that came back? The answer to that second question is where your next revenue stream lives.

Sources

Ingka Group Newsroom --- "AI and Remote Selling bring IKEA design expertise to the many." Ingka Group, 2023. https://www.ingka.com/newsroom/ai-and-remote-selling-bring-ikea-design-expertise-to-the-many/
PYMNTS.com --- "IKEA Uses AI to Transform Call Center Employees Into Interior Design Advisors." PYMNTS, 2023. https://www.pymnts.com/news/retail/2023/ikea-uses-artificial-intelligence-transform-call-center-employees-into-interior-design-advisors/
Inc. Magazine --- "Ikea's Jesper Brodin Thinks AI Is Inspiring--and Deeply Dangerous." Inc., 2024. https://www.inc.com/stephanie-mehta/ikeas-jesper-brodin-thinks-ai-is-inspiring-and-deeply-dangerous.html
Entrepreneur --- "Klarna Is Hiring Customer Service Agents After AI Couldn't Cut It on Calls, According to the Company's CEO." Entrepreneur, 2025. https://www.entrepreneur.com/business-news/klarna-ceo-reverses-course-by-hiring-more-humans-not-ai/491396
MLQ.ai --- "Klarna CEO admits aggressive AI job cuts went too far, starts hiring again after US IPO." MLQ, 2025. https://mlq.ai/news/klarna-ceo-admits-aggressive-ai-job-cuts-went-too-far-starts-hiring-again-after-us-ipo/
Tech.co --- "Klarna Reverses AI Customer Service Replacement." Tech.co, 2025. https://tech.co/news/klarna-reverses-ai-overhaul
Retail Gazette --- "Ikea embraces remote interior design as AI transforms sales tactics." Retail Gazette, 2023. https://www.retailgazette.co.uk/blog/2023/06/ikea-remote-interior-design/