Realising our users are people with actual feelings.
People are complex. This is fact.
In our work as UX designers, we always say that we are building for the end-user. But when it comes down to it, do we really know who our users are?
Yeah sure we can say that one user is a 20-something young professional living and working in London and another is a retiree that likes Tesco artichoke pizza and the TV show Antiques Roadshow but this demographic insight can only tell us so much.
I don’t know about you, but my mood changes drastically from pre to post lunch. I’d be lying if I said good food had no impact on my interpersonal behaviour. My less hangry self is more likely to engage with something that is chirpy and explanatory, especially if it a virtual agent that covers the mundane questions a real-human doesn’t want to be bothered with.
Often times, when we are basing our conversational designs off of our personas, we draw up these caricatures and then subsequently forget that real people aren’t static. Instead of just focusing on descriptive traits, it’s important to recognise that people are volatile beings. We are emotional. We have feelings.
For the HR chatbot we built to support young professionals who were struggling with their career journey, my drawing board started by drawing out who these users were on paper but also in person. In other words, what mood are these users in when they engage with my chatbot? How are they feeling?
We knew who our users were very broadly. They ranged from 22 to 26 years old. Millennials. They were smart and highly capable individuals who had all landed a cushy City job. They were jobseekers during a time when you had companies with perks like beer on tap at the neighbouring WeWork or financial service companies that had higher pay for longer hours and free food.
But within this group of young professionals, there were different kinds of people soliciting help from a careers chatbot. Even if they are looking for the same information, presentation is key and a major make or break on whether the user will be receptive or drop off.
Overlaying the quantitative HR data that showed who was the most disengaged with qualitative insights from my user interviews, we focused on three types of emotional states: the ‘Disengaged’, the ‘Proactive’, and the ‘Worrier’.
The tonality and length of our conversational design depended on the emotions we were responding to. People who are overwhelmed want more encouragement. Ambitious folks want reassurance that they are meeting key performance metrics for promotion and are being given the right trainings to succeed. Apathetic individuals don’t want overly positive messaging that sounds like corporate propaganda.
User testing was our real moment of truth. If the conversational design is correctly informed by the users, the user feedback will validate the insights gleamed from qualitative user research. The chatbot’s responses should fill where the emotional gaps were and address areas where old processes weren’t resonating with users in the first place.
The most satisfying feedback from users is when they compliment the tone of the bot and find the responses really useful. My favourite moment was when your classic ‘Disengaged’ type told me that he actually liked what the bot was saying and wished he had this resource when he first began at the company. He felt the responses were “encouraging and actually useful” rather than just being “a corporate mouth piece”.
Moral of the story? Even if the information is correct, if the way it’s presented is poor and misses the mark, the message is lost. By putting people first, we can have greater intentionality when designing conversations so that human emotions are accounted for and guides the chatbot service experience.
