• HR Intelligence

Building trust: lessons learned from implementing HR chatbots

Great expectations of new tech, project plans, launch day and before you know it, it’s BAU. We’ve been there and got (so) many t-shirts for them. We want to share with you, our lessons learned from implementing HR chatbots. We’re going to tell you about them here so that you can prepare.

Managing user expectations

Dickens was right, there are ‘Great Expectations’. This especially applies to new technology such as AI and automation. The media has created extensive hype. This means there is a disconnect between your users’ opinions of what the chatbot should be able to do and what it can do. This needs careful management.

In our previous blog ‘3 things to think about’, we talked about where the bot fits into your communication channels. As you consider this, you need to also remember:

1. What is your bot is there to do? Help your users understand what the purpose of the bot is along with how they can get the best out of it. This means that they have a useful interaction and are more likely to use it again (yes the ROI writes itself).

2. Be upfront in your communications. We’ve found that language and terminology can create heightened expectations. Users expect ‘AI’ to be much more advanced than it currently is. Does describing it as 'AI' matter? There are other words which help to manage expectations better: automation, computer, bot or simply streamlining or technology efficiency.

3. Think about your key stakeholders beyond the usual senior leaders. Who in your business could help this technology take off? Consider where you may need stronger advocates and how you can do this. Do your people love a giveaway (Bot on a reusable coffee cup, anyone?), do they like a personal story? Engage in different ways.

There’s a lot you can do, but you need to plan and execute.

Consider who your testers are

There is no doubt that you need to test your bot before launch. You need to be confident with how questions are being asked and that the answers are aligning. But, each organisation has its own quirks and terminology. This can mean, to begin with, you might get some quirky answers. Testers who know and understand what the answer should be is more helpful than someone who will simply accept the bot’s answer.

That being said, you can’t have this be too big a group of people. This is definitely one of those times when small and specific is best. Your testers need to be pragmatic but not put off by wrong answers and helping the bot to learn in this critical phase. It happens. The bot gets something wrong in its answer. That’s what testing is for; refining, assuring and moving forward with a better bot. Testers need to be comfortable with these circumstances and understand the critical role they play in helping to get the bot to perform at its best.

Remember your coaching role

It’s not only users who can get swept up in the hype about what the tech can do, but it can happen to you too. Bots are different from any other tech you’ve implemented or used before. An HRIS has certain functionality that doesn’t change and stays the same. It may work for you or it may not, but you’ll know what it is and what it should do. Your bot will not be like this. From the interactions that happen, your bot will be learning, improving and constantly evolving.

It does take patience and attention so remember to manage your own expectations in advance. Think of yourself as having a coaching role with this tech. It’s a part of your team and like any new team member it needs a little time investment to thrive and flourish in the longer term. The more it knows and understands, the more people ask it. It can answer more meaning, longer-term, it frees up more time with greater consistency.

Is There a Perfect HR Chatbot for You?