IKEA’s Billie Chatbot: AI Savings and 8,500 Reskilled Workers

Ingka Group, the largest IKEA retailer, says its AI chatbot Billie saved nearly €13 million by June 2023, roughly $14 million at the exchange rate on the day it announced the figure. The more interesting part is what Ingka did with its people. While Billie handled simpler enquiries, the company said it had reskilled 8,500 call center co-workers, including in remote interior design.

That makes a good corporate headline. The useful question is what the work actually involves.

Watch the table move

In IKEA’s own room film, furniture on wheels makes the same space more flexible. It is a simple idea until that room has to be an office, a dining room and a spare bedroom. A sofa’s dimensions tell you whether it fits against the wall. They don’t tell you whether you can live around it, with someone working from the living room, someone else watching television, a guest who needs a bed and a door that still has to open. For a furniture retailer, helping with that decision could be valuable.

What Billie handled

Ingka, which says it operates 411 IKEA stores and accounts for around 87% of IKEA sales, rolled Billie out in its 2021 financial year. By June 2023, it said Billie had resolved about 47% of the customer enquiries it received between 2021 and 2023.

The savings figure is cumulative and historical, not $14 million of new profit every year. It is also company-reported, not an independently verified return on the investment.

New skills, several roles

Ingka said the 8,500 co-workers built skills in remote interior design, digital sales, relationship building and handling complex customer problems. That does not mean every one of them became a designer. It means the company was building skills for work beyond the routine answer.

Two questions show the difference. Whether a cabinet is available needs a dependable answer. Which cabinet fits an awkward corner may need someone who asks a better question.

A plan and a person

IKEA US launched its Interior Design Service in April 2023, at $99 for consumers. It starts with a questionnaire about the customer’s wish list, followed by three virtual one-on-one sessions with a designer, and produces a mood board, floor plan, 3D renderings and a curated product list. A wish list becomes a plan someone can actually shop from.

The practical choices are where that help earns its place. The dining table also has to be a desk, the sofa has to become a spare bed, and storage can’t swallow the space it is meant to save. The expensive mistake is buying all the right-looking things and discovering they don’t work together.

What a visual planner can do

IKEA Kreativ lets customers place furniture in scale-accurate rooms, including a scan of their own space, and moving a table on screen makes a layout easier to test. It doesn’t show whether that layout suits a household’s everyday life.

It is also a different tool from Billie. The chatbot handles enquiries, Kreativ visualizes the space, and an advisor helps someone work through the decision. The opportunity for IKEA is the combination: moving a customer from browsing furniture to choosing what to buy.

The $100 million distinction

IKEA US reported that its remote channel, covering phone, chat, social media and specialized services, passed $100 million in sales for the first time in its 2025 financial year, and that more than 73,000 customers connected with its planning experts.

Those figures show scale. They don’t show that Billie generated those sales, or that every planning conversation became an order. Sales are not profit either, and the published figures don’t give a full return after software, training and service costs. It is the same distinction behind why AI that works doesn’t always make a business more money.

The limits of the success story

This is not a promise that nobody loses a job. In March 2026, Ingka said 800 roles may become redundant within its group functions as part of a simpler organizational structure. It did not identify Billie or AI as the cause.

What makes the case interesting is the choice to develop people’s skills alongside the automation. A company can answer a question faster and still leave the customer stuck at the next decision. The bigger opportunity may be helping them make it.

The next customer decision

The floor in that crowded room hasn’t grown, but a better plan can make it much more useful. For any business buying AI, the worthwhile question is what its people can help customers do next.

What would you want a real person to help with before buying a room full of furniture?