IKEA 的 AI 策略不只是以聊天機械人取代電話服務
IKEA's AI Story Is More Than a Chatbot Replacing Calls
作為 IKEA 最大零售商的 Ingka Group，其 AI 遙距銷售策略結合 Billie 處理較簡單查詢，並再培訓員工提供專門客戶支援及遙距室內設計。Ingka 稱，Billie 在 2021 至 2023 年間解決約 47% 交予它的客戶查詢，涉及約 3.2 million 次互動，並節省近 €13 million；另有 8,500 名電話客服中心員工接受再培訓。
A widely shared summary of IKEA's AI strategy can sound like a neat equation: AI took nearly half the customer enquiries, workers were retrained, and the company made €1.3 billion in sales.
The numbers are real. The equation needs more care.
The source is Ingka Group, IKEA's largest retailer. Its June 2023 account describes two related parts of a remote-selling strategy: an AI chatbot called Billie, and employees trained for more specialized customer support and remote interior design. Ingka's figures are company-reported, and they refer to different measures and time periods.
What the figures actually say
Ingka said that from 2021 to 2023, Billie resolved approximately 47% of the customer enquiries it received. The company reported this as about 3.2 million interactions and nearly €13 million in savings.
That is not the same as saying AI handled 47% of every IKEA customer's questions, across every country and channel. It is a percentage of enquiries received by Billie, as reported by Ingka.
Ingka also said that 8,500 call-centre co-workers had been reskilled. The skills included remote interior design, digital retail sales, relationship building, and handling enquiries that require more complex problem-solving. The company's description is that automation took on simpler questions while employees could focus on more involved customer interactions.
The third number is €1.3 billion. Ingka reported that sales through its remote customer meeting points reached that amount in FY22, representing 3.3% of its total sales. FY22 ran from September 1, 2021, to August 31, 2022.
That €1.3 billion is not reported as revenue produced by Billie or by the 8,500 reskilled employees. It is a separate sales figure for Ingka's remote customer meeting points. Putting the three numbers in one sentence can imply a causal link that the published figures do not establish.
The more useful lesson
This case is worth discussing because the change was not presented only as a chatbot rollout. The company paired automation of simpler enquiries with a plan to develop workers for other customer-facing work.
That does not prove the transition was painless, that every affected worker moved into a design role, or that the new work was better for every employee. Ingka's public account does not answer those questions. It also does not establish that AI caused the remote-sales revenue to grow.
But it does show why the effect of AI at work should not be measured only by the share of conversations a bot resolves. A fuller account would also ask what happens to the work that remains, whether training leads to real opportunities, how service quality changes, and who benefits from the productivity gains.
For companies adopting AI, the practical question is not simply, “What tasks can we automate?” It is also, “What work will people do next, and are we investing in the skills and support to make that transition real?”
The IKEA example offers a reported strategy, not proof that this model will work everywhere. The numbers support a more limited conclusion: Ingka used Billie to resolve a substantial share of the enquiries routed to it, reported reskilling 8,500 call-centre co-workers, and separately reported €1.3 billion in FY22 remote-meeting-point sales. The relationship among those outcomes should not be overstated.
Sources
來源：DEV Community · dev.to