Conversation Design Best Practices: What Is a Good Conversation Flow?

People working on their laptops, with a pop-up dashboard showing conversation logs.
play Watch

Get a comprehensive intro to conversation design and find out why it matters for your customer satisfaction. Plus, learn the best practices in conversation design from one of our Automation Consultants at Ultimate.

Consumers are increasingly comfortable communicating with bots. The popularity of voice assistants like Siri and Alexa has eased customers into adopting smart technologies. And this growing presence of bots in our daily lives has created a need to make these interactions more useful — even enjoyable. Here’s where conversation design comes in.

What is conversation design?

Conversation design is the process of creating useful dialogues between bots and people. Informed by linguistics, psychology, and technical understanding, good conversation design mirrors natural human speech. The aim of this practice is to develop human-centric interactions with AI-powered bots.

According to PSFK, 74% of customers today would rather use a chatbot to solve simple issues. But some companies are still concerned that automating their customer support will damage the customer experience. They shouldn’t worry. In the customer service space, conversation design is changing the game — and working to humanize automated support.

Why does conversation design matter?

We’ve all had those frustrating bot conversations: where the chat goes round and round in circles without bringing you any closer to an answer. Maybe the bot doesn’t escalate your issue to an agent at the right time. Or maybe it hasn’t understood your question and doesn’t provide an option for you to rephrase. This is a result of poorly designed chatbot conversation flows.

Not only do these interactions damage the customer experience, but they negatively impact your brand. The Conversation Design Institute puts this down to the personification of chatbots. By giving bots human-like features (such as names), customers come to see these bots as employees of a company. And this means they expect human-like levels of intelligence.

This is where you’ll notice the difference between bots. Simple rules-based chatbots can only recognize and respond to preset keywords or phrases. This means they struggle to understand typos, synonyms, abbreviations, or sentences that contain multiple different intents — meaning the reason a customer is making contact with a brand. Because of these limitations, traditional chatbots often use buttons to guide users through a decision-tree conversation flow.

Advanced customer support automation platforms, meanwhile, are enhanced with conversational AI. This allows these AI-powered bots to understand the overall intent of a message without relying on specific words. It also means they can recognize natural human language as it is truly used (mistakes, colloquialisms, and all). VAs use machine learning to become more accurate over time: so the more data they are fed, the smarter they will become.

Although support automation platforms (like Ultimate’s) are far more sophisticated than rules-based bots, they haven’t quite reached human levels of intellect. That’s why thoughtful, well-planned conversation design is needed to create successful human-to-bot conversations.

Best practices for writing better bot scripts

So how do you build dialogues that provide customers with the answers they need? We had a chat with Marika, Lead Automation Consultant at Ultimate. She explained the 3 core elements of writing chatbot conversation flows to keep customers engaged and happy.

1. Know your brand

All customer service scripts should start with your brand — whether for human agents or support bots. Here’s what Marika had to say:

“People ask me what makes a conversational flow great. I start by thinking about the brand persona. This helps me form a better picture of who is communicating, and what their personality and tone of voice is.”

- Marika Sarén, Lead Automation Consultant, Ultimate

Read more tips on how to build an addictive chatbot dialogue flow.

2. Make it natural

Once you’ve decided on the right tone of voice to reflect your brand, it’s time to start planning your dialogues. The key to building good chatbot flows is to keep the phrasing natural. As well as this, it’s important for bot-to-human chats to follow the same logic as human-to-human conversations. So if a customer asks ‘how do I return my order?’ the bot shouldn’t follow this message with ‘we hope you’re enjoying your new items!’

Marika sums up how she goes about this when designing each flow:

“I think of all the different layers that contribute to the user experience. A conversation between bots and humans is based on how people communicate in real life. Personalized exchanges and interactions that flow forward naturally are things we really want to see. Just like humans, AI-powered bots (sometimes known as virtual agents) can listen and understand customers’ issues, respond effectively by providing advice, or even ask follow-up questions for more details from the customer’s side.”

Check out this recap of Marika's conversation design masterclass for even more expert advice.

3. Plan for the unexpected

“As with human-to-human conversations, bot-to-human conversations don’t always go as planned. Users can break the pre-designed flows by asking questions that are outside of your initial scope.”

But as Marika explains, in these critical places — if you design your flows effectively — you can make or break the customer experience. There are plenty of things you can do to keep customers from going round in circles or hitting a dead end. This might include building in loops to request additional information that will drive the conversation forward. Or it might mean creating different flows for customers who are logged in and those who are not.

Generative AI — and the future of conversation design

The next generation of AI technology, generative AI, has shaken up the support automation space. And this tech is changing how we think about conversation design.

In fact, with gen AI you don't need to design conversations at all. Simply connect the bot to your knowledge base, let it analyze your help center articles, and start having natural-sounding conversations with customers in minutes. Instead of using pre-designed flows (which take time to set up) the bot can draw on its extensive training data to understand and summarize text, and generate accurate answers to your customer's questions.

Read this article on how to build an AI chatbot with ChatGPT.

But generative AI doesn't mean conversation design is dead.

Far from it. While generative AI helps you get to instantly start deflecting FAQs, pre-designed conversation flows allow you to automate complex processes end to end. The best automation solutions combine both approaches, bringing you the best of both.

When thinking about automation, it’s important to choose a provider that gives you the tools to design on-brand, customer-centric conversation flows — where needed. And the power to deliver most human-like conversational experience with generative AI (for taking care of FAQs and chit-chat). Combine world-class AI technology with an understanding of conversation design, and what do you get? Automated customer support that really enhances CX.

Want to learn more about conversation design?

Watch our free digital event