Frustrated customers have long dreamt of seamless conversations with smart chatbots rather than the often disappointing process of going through a call center to find help. Now, advances in large language models (LLMs) are bringing that prospect closer to reality and enabling businesses to deliver the level of experience customers demand.
However, while implementing conversational artificial intelligence (AI) may sound like a silver-bullet solution, chatbots take time and human effort to implement effectively. Furthermore, poorly designed chatbots can pose significant risks, especially for companies in highly litigious or regulated industries.
We explored how businesses can bring chatbots in line with customer expectations by speaking to Rachel Whitehorn, Conversation Designer & Lead CRM SME Strategist at GEICO, in the latest episode of our Conversations That Matter podcast.
Want to Learn More About How Businesses Can Bring Chatbots in Line With Customer Expectations?
Check out our discussion with Rachel Whitehorn on Conversations That Matter.
Myth Debunked: Chatbots Are Easy
As is traditional in our podcast episodes, we began by asking Rachel to debunk a popular myth. She told us that many companies were excited to get a chatbot during the COVID-19 outbreak as it seemed a quick and easy solution to their call center challenges.
But, as Rachel explained, it’s not that easy: “You can’t just plug in a chatbot and expect everything to be solved. Many organizations that thought they could do this quickly didn’t involve all the necessary departments or people, and many didn’t even have conversation designers.
“Chatbots are good virtual assistants, but they have to be done in the right way to benefit you; otherwise, you’re risking your brand. Because if the experience isn’t what customers were expecting, then you’ve let everybody down.”
Intentional Design Is Critical To Delivering On Chatbot Expectations
Amid AI’s recent heightened attention, Rachel says customers’ expectations aren’t changing hugely. That’s because customers often don’t know how good a bot is; they simply want it to help them achieve things more quickly than if they were speaking to a call agent and are left disappointed when it doesn’t.
The risk is higher in industries like finance and insurance, so the onus is on other sectors to lead the way and prove chatbots’ worth. Rachel told us: “People are going to need to be brave. The less litigious industries will need to go first when it comes to integrating LLMs and discovering what can go wrong. The key is being intentional in the way you design, and all these companies that are interested in having virtual assistants go well need to be intentional.”
To deliver on this potential, people will need to embrace change. That means conversational designers will need to adapt to the new technology, which Rachel believes will see their roles merge with those of AI trainers.
Getting Started With Conversational Design
Rachel said: “When people are looking for help, and they see the little icon in the bottom right, many of them will click it. Having bots go the right way can be difficult because customers like being guided but hate being railroaded. With a lot of bots, it feels like we’re railroading them to communicate in a certain way to get what they want, and users don’t like that. That’s another thing I think is going to come from LLMs — not having to railroad them as much to get what they want.”
Rachel also believes the onus is on conversational designers to learn how AI can solve customer problems. She explained: “What I’m afraid of for some designers is that they box themselves into ‘this is all I do.’ They have to be able to wear different hats; so speak to stakeholders about your results, automation, and containment, and you can become an advocate for the bot. You have to know more than just design to really function at your best.”
The Future Of Self-Service Experiences
AI technology is evolving rapidly, and Rachel is hopeful that significant steps will be made in the next few years: “I’m a dreamer about how quickly tech can meet my expectations. But what I want to see within the next two to three years is making generative AI so that the conversation designer is curating responses, not creating responses. My dream is for the bot to report to me that it’s answering so many questions without a lot of guidance and working from a framework of intents, but when it goes off path, it doesn’t have to say, ‘I don’t understand.’
“If there’s a high volume of questions because something changed, I can teach it what the company line should be; then the bot should be a true assistant. Then at the end of the week, I want it to report to me saying, ‘These are the things I wasn’t so confident about, can you help me?.’ Then I can guide it on what it should have said to help answer the user’s question.”
Maximize Your Customers’ Chatbot Experiences
As AI chatbot capabilities evolve rapidly, customers will begin to expect even greater experiences from businesses. That means companies must take action to implement intentional design and start delivering on customer expectations.