What EY Learned When It Started Using HR Bots

Article main image
Sep 10, 2018
This article is part of a series called Editor's Pick.

Technology has played a significant role in shaping human resources practices in the past decade. Today we are on the cusp of a new revolution in HR: The application of artificial intelligence and machine learning toward HR service delivery. In short, there is a lot we don’t know about how HR will evolve in the coming years. But we do know one thing for sure: The bots are coming.

Chatbots, typically built on a foundation of machine learning that is consumed as a personalized, conversational interaction between a person and technology, have been in use for some time in other areas, like customer support and personal devices (think Siri). But these tools are making their way inside enterprise walls in a big way and HR is well positioned to lead their companies’ transformation from human-led service delivery and processes to efficient, effective ways of linking people and tools to get things done.

Most HR professionals are aware of bots. But it can be very difficult to know how to incorporate this new and often confusing technology into an HR strategy, or even into day-to-day operations. Are bots just cool and innovative, or can they really be applied to solve business challenges? What impact does increasing people’s reliance on technology have in today’s workforce, full of people who desire more personalized experiences and human connection at work?

Penny Stoker, Global Leader of HR Services at Ernst & Young, was asking herself these same questions when EY embarked on an exploratory project to incorporate chatbots into their HR service delivery model. EY’s journey helped us outline some basic dos (and some definite don’ts!) when it comes to exploring, leveraging, and incorporating chatbots in HR.

Don’t adopt a bot because it’s cool and new

It is tempting for organizations to adopt cutting-edge technology just because it’s cutting edge, especially given that outdated technology is diminishing the employee experience, reducing engagement, and hurting the bottom line. But the most impactful enterprise technology is that which helps make people’s lives easier in some way.

Do identify the use case or cases

“For us, we started with onboarding, where a lot of processes and information can be repetitive from one new hire to the next,” Penny shared. EY’s Onboarding Buddy was designed to give new hires critical information that was previously delivered by HR professionals, as well as answer questions throughout the onboarding process. Other typical use cases range from the very simple (automated self-services, like changing one’s address or awarding a spot bonus) to the complex (intelligent recommendations for what learning course someone should take next, or what team member is best suited for an upcoming promotion).

Where are there bottlenecks to employees or managers getting the information they need? What are the administrative things HR professionals must get through in order to focus on more strategic issues? These questions can serve as the beginning of an investigation into where a bot might add the most value.

Don’t be intimidated by the technology

A chatbot takes machine learning insights that change and develop over time, and turns these into a conversation. If it sounds complicated, it is! But if your organization waits to develop the expertise needed to understand all the underlying processes, you’ll never get to the point where you’re ready to incorporate the tool.

Do find the passionate people

“We didn’t have anybody on our team that was a total expert on chatbots or AI,” Penny told us. “What we did have was a lot of interested people who were really passionate about figuring out what this technology could mean for our people.”

Today, it is highly unlikely that your HR organization will need to build a chatbot from scratch. (But you can if you want. See “Simplify Your Social Media Recruitment By Building Your Own Chatbot.”)

Many technology companies including my company, SAP SuccessFactors, are starting to deliver this technology to organizations. Since the underlying technology is already there, the real work to be done is around training the machine and having it learn your people’s needs.

“We were very clear with the workforce from the very beginning that they all play a role in helping this tool become successful,” Penny said. “In that way, everyone was able to buy in and feel a part of the process. The focus was less on ‘How does this work technologically?’ and more on ‘How can I help increase accuracy through feedback I give to the tool?’ And since that feedback is administered really simply and conversationally, this mindset greatly reduced fear and intimidation.”

Don’t expect perfection

Because chatbots are built on a foundation of machine learning, they have to do exactly that — learn.

“What differentiates chatbots from other technologies is that they’re not going to be perfect out of the box,” Penny said. “It’s called machine learning because the tool learns over time. One of our big focus areas in rolling this out was making sure this was clearly communicated.”

Do test and learn and share successes

Engaging the workforce to help train the tool was critical. This involved testing, and learning from the different ways people interacted with the solution.

“Millennials were more likely to interact with the chatbot like it was an actual human,” Penny shared. “They were more likely to ask fun questions like, ‘Are you married?’ It was a really fun exploration into how different segments of people interact with technology when they’re given the chance.”

In addition to watching the chatbot gain accuracy over time via its interactions with people, EY HR leaders were able to observe some additional interesting tendencies among their workforce.

“We did see increased accuracy in the information the tool was able to provide to people,” Penny said. “But one of the things we discovered was that people were far more likely to give the tool negative feedback — indicate when something was incorrect or not helpful — than they were to give positive feedback. The tool, like the typical employee, needs that positive feedback to improve as well.” This lesson helped inform new communication strategies around the tool.

There was one specific point at which Penny and her team realized how powerful the chatbot technology could be. When asked a question about a policy, the chatbot returned information to the HR service delivery team that seemed incorrect. But in actuality, the policy had recently been changed, and the bot picked up on that change; it was the team that had it wrong.

“That’s when we realized how incredibly useful something like this can be,” Penny said. “Rather than have human beings pore over policy documents when they could be strategically engaging the workforce, organizations can leverage bots to guide people in their decision-making with full knowledge of the policies and what they mean—instantly.”

This article is part of a series called Editor's Pick.