Human-centred Conversation Design: The Secret Sauce Powering Better AI-enabled CX


Artificial intelligence is woven into the fabric of modern life. It decides which tweets appear on your Twitter feed and recommends YouTube videos for you to watch. AI is involved when Google adjusts your route to follow the fastest way home or shows an ad for the Bluetooth headphones you’ve been dreaming about.

The popularity of AI-powered voice assistants like Alexa, Siri, and Google Assistant means people today are more comfortable interacting with human-like computers. PSFK research found that 74 per cent of people prefer using chatbots to answer simple questions. But chatbots still have a bad reputation, especially regarding customer support. We’ve all been burned by frustrating bot experiences where you’re stuck in a loop and can’t get the help you need.

Many businesses turn to customer service automation to reduce costs and increase efficiency. But conversational AI technology advances mean support bots can do more than basic deflection. In fact, AI-powered chatbots can handle complex conversations that deliver experiences that feel empathic and human. The key is to keep customers at the centre of every automated conversation.

Why human-centred conversation design matters

Accenture predicts that 95 per cent of customer interactions will be AI-enabled by 2025. And while customers are happy to use a chatbot to get the answers they need (particularly if it’ll help solve their issue faster), they’re not willing to sacrifice superior CX.

Support teams at the most customer-centric companies know that customers’ expectations are higher than ever — and they’re willing to look elsewhere if businesses miss the mark. 50 per cent of customers will seek out a competitor after a single bad experience. This means brands must be sure their bot can deliver the same joyful CX as their human agents.

Here are three ways human-centred automation can help companies deliver outstanding CX (and the business benefits you’ll feel).

  1. Start with empathy

    There’s a story behind every support ticket. That’s why when customers reach out to support teams, they’re not just looking to have their questions answered. They want connection and reassurance — as well as fast, accurate resolutions.

Hans van Dam is the CEO and Co-founder of the Conversation Design Institute, a leading organisation that offers world-class training in conversation design skills. Hans explains that when customers ask for support unlocking their bank account, they might also be saying they couldn’t pay rent or buy groceries. “We need to recognise that as creators, otherwise, we just build cold machines.”

Human-centred design starts with empathy, and empathy comes from understanding. So before your team builds any automated dialogue flows, get to know your customers: how they speak, what’s important to them, why they reach out to your brand and the channels they use.

Truly understanding your customers — and their pain points — is essential to delivering empathic automated experiences. Then, once you start designing conversations, make sure your customers feel heard and understood. Have your bot repeat a customer’s concern back to them (“I’m sorry you’re having trouble accessing your account”) and show that you care (“that must be frustrating”).

Your bottom line will thank you, too. By creating AI-enabled interactions that keep customers coming back again and again, you’ll drive loyalty and grow revenue.

  1. Develop a learner mindset

One of the key principles of human-centred design is being open to learning. So only assume you know all the answers after validating the data.

Before starting to automate, read through past support conversations to find out the most common questions, your customers are asking. The best automation providers can leverage AI to analyse this historical support data to determine what your customers are asking most often. Many CS teams will be surprised that their existing FAQs page doesn’t offer customers the answers they want.

Most conversational AI platforms also provide analytics and reporting dashboards. This allows you to measure key metrics like bot handled rate (the number of tickets resolved without a human agent’s involvement), message understood rate (which tracks the accuracy of your bot’s responses) and your top intents (the questions customers are asking most often). Armed with this data, you can train your bot to become more accurate over time, deliver more personalised CX, and drive customer loyalty.

  1. Test, test, and keep testing

Human-centred automation shouldn’t be a set-and-forget system. To deliver the best experience for customers, you have to keep iterating.

Before you launch a new AI-powered bot or conversation flow, make sure you test it thoroughly. Are there any loops that customers could get trapped in? Is the choice of language on brand? Have you included smart escalation paths?

After launching, make sure there are opportunities for customers to give feedback. And keep training your bot — the beauty of AI is that every automated interaction will improve over time as the AI model learns and develops. This means customers receive more accurate support that instantly solves their issues, resulting in a better experience.

To sum up

When designing AI-enabled experiences, it’s essential to remember why automation was created first: to take over repetitive tasks and give people time to focus on what truly matters.

Human-centred design is all about customer-centricity. So using this framework when creating automated interactions ensures peoples’ (specifically, customers’) needs aren’t sidelined by efficiency and cost-cutting concerns. By taking a human-centred approach to automation, you’ll deliver faster, better CX — as well as provide customers with the connection they crave.

If you liked reading this, you might like our other stories
Does Fragmentation Impact Your Omnichannel CX?
The Impact Of Speech Tech On CX