zapp2photo - stock.adobe.com
Before deploying recruitment chatbots, define use cases
HR departments can benefit from using chatbots for basic tasks, but the technology will eventually be able to handle more complex requests. Be sure to set clear expectations first.
L'Oreal Group fields about 2 million job applications a year for its open retail positions and internships. For the makeup giant's HR team, that means answering numerous questions and offering frequent assistance, not to mention making a lot of tough decisions.
Luckily, however, they no longer have to do this all themselves. For the past year, L'Oreal's retail and internship hiring efforts in France, the U.K. and the U.S. have been backed by AI-powered recruitment chatbots designed to help candidates get answers to questions and build their profiles. The company can then analyze the resulting data for insights to help with hiring decisions.
In that time, Mya Systems, the vendor that built and deployed L'Oreal's chatbot technology, has surveyed 10,000 candidates, and 99% have said they were "very satisfied" with the chatbot experience.
L'Oreal's recruitment chatbot experience is definitely not typical for a technology category that's still in its relative infancy, but it serves as notice that, when done right, chatbots are already capable of delivering significant impact to a company's recruitment experience and processes.
According to Eyal Grayevsky, CEO and co-founder of Mya Systems, based in San Francisco, simply installing a recruitment chatbot and expecting it to solve problems is inviting disaster. For a chatbot to succeed, a lot of consideration must be given to how it will be used, what it will be doing and what's likely to go wrong.
Grayevsky suggested taking steps such as escalating certain high-value content, or providing guardrails for dealing with moments when the chatbot doesn't have an answer or the user says something unintelligible.
"It's really important to design the product, in addition to deploying the technology," Grayevsky said. "There's an art and a science to this stuff."
Improving the candidate experience
Mya's technology was born from an early hypothesis that job candidates don't hear back from prospective employers often, don't have a forum to really communicate their value and don't have visibility into the process.
David KarandishCEO, Capacity
To tackle this, Mya has built a system that has over 3,500 intents that are specific to the recruiting domain. Each question asked of a chatbot has intent tied to it. If the answer is a simple yes, that's one intent. If the answer is something like, "Yes, but I can't work on Sundays," then that's two intents.
As the list of intents it supports has grown, Mya has expanded into additional use cases, such as interview scheduling and calendar management. It has also started experimenting with outreach automation for times when it needs to engage passive candidates.
The company recently posted its 25 millionth interaction with a job candidate, and that has given Mya lots of data with which to work. That data is continuously looped back into the product, and human annotators chime in when there are gaps. All of this helps to ensure Mya's chatbot technology can deliver the connection with and visibility into the recruitment process that candidates crave.
The approach appears to be working, as Grayevsky said Mya recently conducted a comparison of users applying for light industrial and call center roles, and 90% of those who used its chatbot technology completed their application processes -- a higher rate than those who did not have access to the chatbot.
Closing the feedback loop
Capacity, formerly known as Jane.ai, also relies on a feedback loop for those times when the answers available to the chatbot simply don't match a user's question. The company has designed its chatbot technology to first ensure it's responding to the right question if there's any confusion. It presents its interpretation with a qualifier -- "Here's what we thought you asked," for example -- so the user has some idea of what's going on. Capacity then relies on its feedback mechanism to fill in any gaps.
Think of it all as deep learning on the fly. David Karandish, Capacity's CEO, said one of the reasons chatbot technology has such promise is the fact that so many companies have moved their CRM, ERP and HR systems to the cloud. Karandish argued cloud providers like Salesforce and Workday are "really good databases in the sky," but the way users engage with that data leaves much to be desired.
"The reason these chatbots exist," he said, "is that companies have to have a system of engagement to act as a gatekeeper."
As such, recruitment chatbots today are effective at straightforward tasks, such as answering frequently asked questions, onboarding new people -- employees, customers and partners -- or retrieving answers from documents. But they will be capable of more all the time, and Karandish said he believes the technology is already providing a recruiting advantage that will only get bigger.
"If you're not investing in AI and chat technology, there's a real opportunity to be left in the dust behind your competition," he said.
John Sumser, analyst and editor in chief of HR Examiner, agreed that even though chatbot technology is so nascent, the current best practice is to simply try something so the organization can learn and, ultimately, move a step ahead of its rivals. For that reason, Sumser never tells clients to wait on using chatbots, but he does encourage them to have realistic expectations at first and to be very clear why they are doing it.
"The benefit is that the chatbot will show all the holes in your thinking about what candidates want to know about your operation," Sumser said. "It's a great way to get on the road to fleshing out all the things you need to do in order to become a 21st-century company."