30,000 examples of how Ikea works with AI

Only a few years ago, Ikea developed a group-wide data effort with a particular focus on AI to manage investments, and it’s been a focal imperative ever since. 

“If followed right, more data can be used in the apps and AI solutions that are built as new data comes in,” says Francesco Marzoni, chief data and analytics officer at Ingka, the holding company that runs most of Ikea’s department stores. 

Marzoni is in charge of a department with upward of 500 people who work with data analysis and machine learning (ML). And then there’s also a traditional IT department with system architects and developers. 

“We work together in cross-functional teams to develop different solutions, and our collaboration works very well,” he says. “If managers on the technology side were arguing, we’d really have a problem, but that’s not the case.” 

One-year, full-time education

The expertise around AI and data isn’t isolated in Marzoni’s department. It really reaches out into the business, and Ikea has started a program to capture talent internally through one-year education in data analysis where 10 to 15 people study full-time.

“It’s about people from sales, supply chain, and HR, for instance, who are allowed to engage in data analysis for 12 months,” he says. “They must then take this knowledge with them to their parts of the company.” And interest in the training has exceeded expectations with 150 people already applied. 

“It’s a good sign,” he adds. “It means people want to get involved and invest in their development, and that’s important.” Training in-house is a strategy that Marzoni clearly believes in since expecting people who are experts in a certain area to step in and make things work with the company’s processes right away isn’t realistic at the level they need. Also, AI in particular is a discipline that works in a very specific way. 

“To build AI capabilities that can help employees in a certain processes can be as important as a data scientist or a machine learning engineer being the expert and building the solution,” he says. “AI is very much about data, and data is largely in people’s heads, so we must invest in developing the people — the experts — who become important actors when it comes to how we build solutions.”

Order of data

Having your data in order is the cornerstone to moving forward with AI. “The ecosystem is constantly producing more data, so it’s impossible for everything to be in perfect order all the time; it’s something you have to constantly work at. And it’s important to have your data in order even if you don’t know what you’re going to use it for. It’s very easy to do relevant things with AI if you have a handle on your data. So far, Ikea has come a long way with its data management concerning supply chain and warehousing, but when it comes to customer experience, however, there’s more to do, says Marzoni. “We can improve how we collect data about the customer process,” he says. 

Better control over internal data

The question remains if it’s easier to organize data about delivery and inventory than customers, but, according to Marzoni, it leans more to the former since that’s where they started. “That’s where we saw the greatest potential,” he says. 

Another aspect is there’s better control over internal data than external data, which may be the case with customers. 

“When it comes to customer data, there’s also more to be done in terms of handling it responsibly,” Marzoni adds, “making sure information is provided on how it’s handled. But it’s also an area where we’re strong.” 

More classic AI than generative

The AI ​​solutions used by Ikea today are mainly based on classic ML; the LLMs aren’t used much other than in finished products such as Copilot. For example, AI is used for things like recommendation services or to optimize filling a long-haul truck. And it’s about developing and training your own AI solutions precisely for the areas where it’s relevant. 

“AI for tools such as Office can be bought directly off the shelf, but I think the best AI solutions for healthcare, for example, may come from a player in healthcare,” says Marzoni. “And it’s perhaps more likely that solutions for interior design will come from someone who’s actually doing it and not a big software company. The recommendation tools that Ikea uses also have to be trained internally, and we can then build interior design services into them.”

AI as an interior designer

So an area that Ikea is investing in going forward is digital services for interior design. Not everyone can afford to hire an interior designer, so in this way, people can take advantage of the AI expertise they have.

“It’ll be felt in a lot of different ways,” he says. “Things like waiting times will be shorter and you’ll get help faster. And you’ll get more access to things you didn’t have access to before.” 

AI will have a clear impact internally as well, not least because employees will receive more support with administrative tasks, and more time to devote to basic tasks. “Everything will go faster,” he adds. “So instead of filling out forms, they’ll have more time to give advice in the shop or build kitchens.” 

Strength in numbers

A large proportion of employees, about 30,000, have also received basic AI training since the summer. “This isn’t something we had to push for since there’s a great desire for it,” Marzoni says. In addition to more basic web training, other initiatives that were not planned at the beginning have also emerged, such as hackathons. 

Ikea also has a format called World Data Games where data collected can be used at the Ikea Foundation, which centers around social issues, climate change, among other things. “By analyzing this data, their AI competence increases,” he says. “I’d encourage others to give employees a week for this kind of initiative to successfully move forward in AI.” 

Another important part of this type of initiative is that experts and practitioners work together so there’s a bridge between them. “This means practitioners in the business get direct access to the AI ​​expertise,” he says.