IKEA's AI story is often summarized as a $1.4 billion chatbot success. The reporting behind that headline is more nuanced: Reuters reported that Ingka Group's remote interior-design channel generated €1.3 billion (about $1.40 billion) in revenue in fiscal year 2022, while the company's AI assistant, Billie, handled a share of routine call-center queries. The reported revenue belongs to the broader remote interior-design channel; Reuters did not attribute it to the chatbot alone.
The useful lesson is not that AI guarantees a particular financial result. Organizations should consider how automation, service design, and people's work fit together, then evaluate each contribution separately.
What Reuters Reported
In its June 2023 report, Reuters described Ingka Group's use of its AI assistant Billie and its remote interior-design service. Ingka said it had trained 8,500 call-center workers as interior-design advisers since 2021, while Billie had handled 47% of call-center customer queries over the prior two years. Reuters also reported that sales by phone or video through Ingka's remote interior-design channel accounted for €1.3 billion (about $1.40 billion) in fiscal-year 2022 revenue.
The article reported those figures as separate elements of a wider service strategy. It does not show that the chatbot alone generated the channel's revenue, nor does one company's reported experience establish what another organization can expect.
Reuters also reported that, when asked whether increased AI use was likely to reduce headcount, Ingka's global people and culture manager said, “That's not what we're seeing right now.” This is a statement about the company's view at that time, not a general guarantee about future staffing or the effects of automation.
Our Reaction: Ask What Work Should Change
A more useful starting question than “How many people can this tool replace?” is: Which tasks are suitable for automation, and what work should people be able to focus on?
That question does not assume that automation always benefits employees or customers. It asks leaders to make explicit choices about work design, staff involvement, oversight, service quality, and the use of any capacity created.
Automate Tasks Without Automating Accountability
Some routine activities may be suitable for software assistance, such as organizing information, retrieving approved material, preparing a draft, or routing a request. Whether a specific task is appropriate depends on its context and consequences.
People should have a clear role in checking outputs, handling exceptions, and taking responsibility for decisions that affect customers, employees, applicants, residents, or community members. The tool's role and limits should be understandable to the people who use it and those affected by it.
Make Workforce Planning Part of the Change
Introducing a tool can change the tasks people perform. Organizations should involve affected staff early, explain what is changing, and plan for training, new responsibilities, and escalation paths. A responsible plan should consider:
- Which tasks may change and which should remain person-led?
- What skills and support will employees need?
- How will staff raise concerns or correct system errors?
- How will managers monitor service quality and employee experience?
- What will happen to capacity if routine work is reduced?
Training is not only about operating a product. Staff may need guidance on checking outputs, protecting information, recognizing limitations, and knowing when to seek help.
Decide How to Evaluate the Change
Time saved is not, by itself, proof of improved service or value. Before a change, identify the baseline and measures that reflect the organization's purpose. Depending on the workflow, those may include accuracy, turnaround time, user experience, unresolved exceptions, staff workload, or access to service.
Review results and unintended effects before expanding. If the system shifts work to another team, creates new errors, or makes service less accessible, those findings should influence whether and how it is used.
Start With the Workflow, Not the Tool
Before selecting technology, understand:
- What requests or tasks arise, and how are they handled today?
- Which steps are repetitive, and which require context or discretion?
- Where do delays, errors, or handoff problems occur?
- What information is needed, and is it appropriate to use with a proposed tool?
- When should a person review, correct, or take over?
- What would count as a successful and acceptable outcome?
Automating an unclear or ineffective process can preserve its problems or move them elsewhere. A defined process and clear responsibility make it easier to assess whether technology is appropriate.
What Mission-Driven and Public-Service Organizations Can Take From It
Organizations in government, nonprofits, and business have different goals, resources, and obligations. They should not assume IKEA's operating model transfers directly. They can, however, ask similar design questions: what routine work might be assisted, what requires human judgment, how employees will be supported, and how the service will be evaluated.
Hypothetical example
A public-facing team might explore whether a tool can organize routine questions for staff review, while keeping unusual or sensitive matters on an established human-led path. The team would define approved information, test the process in a limited setting, and review accuracy and service effects before considering any wider use. This is an illustrative scenario, not a Gallant client project or a report of measured results.
A Human-Centered Review Checklist
- Identify the task. Describe the workflow and its users before considering a solution.
- Set boundaries. Specify what the system may assist with and what remains a human responsibility.
- Involve affected people. Include staff who do the work and people who use the service.
- Prepare and test. Provide training, review outputs, and plan for exceptions and fallback.
- Evaluate and adjust. Measure service and workforce effects; expand only when evidence and safeguards support it.
The Bigger Lesson
The reported IKEA figures provide a case to examine, not a promise or a recipe. The remote interior-design channel's revenue, the chatbot's handling of customer queries, and workforce training are related parts of a reported strategy, but they are not interchangeable measures of AI impact.
Leaders should be specific about what a tool does, what outcomes are observed, and who benefits from a change. AI can be one part of a redesigned service; it does not replace accountability, workforce planning, or evidence.
If your organization wants to discuss practical automation opportunities while keeping people and service outcomes in view, book a discovery session with Gallant Business Solutions →.
Frequently Asked Questions
Did IKEA make $1.4 billion solely because of AI?
No. Reuters reported that Ingka Group's remote interior-design channel generated €1.3 billion, about $1.40 billion, in fiscal year 2022. The report separately described Billie handling a share of customer queries. It did not attribute the channel's revenue to the chatbot alone.
Did IKEA fire employees after deploying its AI chatbot?
Reuters reported in June 2023 that Ingka had trained 8,500 call-center workers as interior-design advisers and quoted an Ingka leader saying that a reduction in headcount was not what the company was seeing at that time. That dated statement should not be generalized into a guarantee about staffing outcomes.
What does human-centered AI implementation mean?
It means making the effects on people part of the design and evaluation: involving affected staff and users, defining appropriate limits and human review, preparing people for changed work, and monitoring service outcomes.
How can a smaller organization apply the lesson?
Start with a bounded workflow, specify where human review is required, decide how staff and users will be affected, and test whether the change improves the outcomes that matter before expanding it.
Should AI replace customer-service or public-service employees?
Replacement should not be presumed to be the goal. Organizations should evaluate whether a particular task can be safely assisted, what responsibilities remain with people, and how the effects on service and staff will be handled.