Main content

Workday Rising EMEA - Mondelez International tests out generative AI for employee self-service

Phil Wainewright Profile picture for user pwainewright December 1, 2023
Snack food giant Mondelez International rolled out Workday to 100,000 global employees three years ago - but self-service took longer to get right. Now it's testing generative AI for the next phase.

Mondelez product portfolio © Mondelez International
(© Mondelez International)

With 100,000 employees worldwide, snack food giant Mondelez International is on a constant quest to enhance its HR processes. The US-based owner of much-loved biscuit, chocolate and snacking brands, including BelVita, Cadbury, Clif, Milka, Oreo, Philadelphia, Ritz, Toblerone and many others, rolled out a Workday HCM system across more than 80 countries in September 2020. An important goal has been enabling employees and managers to self-serve simple tasks instead of having to call up the HR team — but it's taken time to get it working well.

Three years later, 90% of employees now self-serve for tasks where it's offered, but can still face issues when attempting something unfamiliar. So the company has been testing whether generative AI can power a more conversational interface for these less frequent interactions. Volker Schrank, Senior Director, HR Technology & Employee Experience at Mondelez, says:

A lot of our workforce is doing a transaction once, maybe twice a year. You can have the simplest of visual interface. [But] if you do it only once a year, you struggle, you need to find your way. But if you do in an interview basis, and a question-and-answer perspective, that is more easy to do that.

The biggest challenge for occasional users is that they're not familiar with the terminology that the system expects, which makes it hard to actually find the self-service task they need. Adding an AI-powered conversational layer means they can ask for what they want in their own words, and the AI can then figure out what they're trying to do, and ask them for whatever information the system needs to get it done. He explains:

For simple applications — I got married, I have a new address, I had a kid, or I need a letter of employment. These are things that are very easy, very straightforward. If you're going to have a conversational interface where they can just use their language, then it works.

We are trying to pull out the HR lingo from all of our applications as well. But in the end, the application is the bottleneck. If you use different wording for what you are searching for, you will not find it ... If you don't find it, because you call it differently, you are lost on the visual interface. You are not [lost] on the chat. This is why we are looking at the chat interface going forward as being the interface for quick, self-service transactions.

Serving a global workforce

The AI interface is also able to operate in many different languages, which is hugely valuable for an organization like Mondelez that operates in so many different countries. Since Mondelez started running tests a year ago, the technology has improved significantly, to the point where it may soon be ready for production use. Schrank says:

The quality you get now is so much better, that certain use cases can be delivered very well. For everything else, the only thing you need is a good hook to then go to a human agent.

This is how we are building it out. Making sure that we are focusing on the 80% of cases that everyone wants to solve. For those, we are trying to optimize pathways. And for the others, having easy access to the human when they realize they need something different.

One advantage of all the hype around generative AI is that employees are eager to be involved in trying out the technology and to give feedback on the issues they encounter. He says:

Almost everyone is currently so excited about generative AI that they are forgiving us. In the past, they were not forgiving. 'You want me to talk to a chatbot? A dumb chatbot? I don't want to.' Now, it's like, 'Oh, we have ChatGPT in the company? Let me try it out!'

The Workday system had replaced an earlier SAP HCM system, along with Oracle Taleo recruiting, SAP SuccessFactors talent management, and ADP Workscape for compensation. It was a 'big bang' implementation, which brought all modules live at the same time across the world — full HCM, recruiting, talent, learning, optimization, advanced compensation, and planning. While the previous systems had been used globally, there were separate instances that were not all used in the same way, so data standardization had formed a big part of the preparatory work, along with integrations to other systems such as benefits. Workday has now become the core system of record for all employee data, including identity. Schrank says:

It's the only system of record, the only talent management system we have. It is also the backbone of the rest of the company, if you will, because any employee data are sourced from within Workday when there are disparate other systems that need it.

Making a success of self-service

But while the back-office implementation went smoothly, the employee-facing self-service capabilities were not initially a success. Schrank explains:

When we designed it during the project, it was HR people who designed it. And when we went live, the managers and the employees said, 'I have no idea what you want from me. I don't know what this word means. I don't know how this system works. I don't know, I [can't] do this.'

This led to a rethink, with much more of a design thinking approach — starting from what people wanted to achieve and then building the processes that would best support those goals. This was mroe successful, as he explains:

We turned it around from being, at the beginning, one of the systems that was most criticized being self-service, to, as of today, around 90% self-service adoption [where offered], across all of our communities.

With around half of the workforce in manufacturing facilities rather than being office-based, the self-service experience also had to be delivered to workers who were not familiar with traditional computing devices. This meant that the first kiosk devices, equipped with a keyboard and mouse, were not successful, and so touch screen devices were put in instead. Employees can also use their own mobile devices to access Workday, if they wish to. This is particularly popular outside of Europe and North America.

Demographics play a part too. Self-service take-up is highest in Asia Pacific and Latin America, where the uptake is as high as 94-95%, whereas in Europe it's around the mid-80s. This is down to differences in culture and the average age of workers between these regions, says Schrank. The hope is that the move to a more conversational interface, enabled by AI, will make self-service more appealing to those who currently still pick up the phone to HR.

A grey colored placeholder image