Best Practices for Creating a Meaningful and Engaging Automated Customer Service Experience with Conversational Design
Advertising copy packs a punch. Online self-help wikis and FAQs offer factual explanations. Conversational AI copy is somewhere in between… and therein lies the challenge of doing good conversational design.
Creating copy for a customer service AI is different than what your company has created in the past. It’s not just about being creative and striking an emotional chord as it needs to answer your customer’s questions quickly. It also needs to engage your customer in a conversation and not be robotic, simply serving up long-form content. It’s a delicate balance.
Here are our best practices to keep in mind as you embark on conversational design for your customer service AI:
1. Keep It Simple
Use short, concise sentences. Long-form paragraphs are hard to read in a chat interface. If you need to provide a lot of information, break it up into multiple messages. Focus on keeping the character count in each message low. Just think about how you text your friends; it’s likely done using a few words, in short bursts.
2. Ask Questions
While the majority of the company and consumer interaction is one-way (ads, email marketing, website, etc.), conversational AI needs to be two-way. Ask customers for input in order to serve up only the most relevant content (this also helps to keep messages short). For example, if your return policy varies based on whether a product is personalized or not, ask your customer if they have a customized product before giving up your lengthy policy.
- Example 1:
- Customer – Can I return a pillow that I just purchased?
- AI Agent – Was your pillow personalized with a name or initials?
- Customer – No
- AI Agent – We accept returns within 30 days of purchase. Please bring the item into any retail store.
- Example 2:
- Customer – Can I return a pillow that I just purchased?
- AI Agent – You can return non-personalized items within 30 days of purchase at any retail store. We cannot accept returns for products that have been personalized.
As you can see, in the first example, the AI asks a simple question in order to serve up only the most relevant content. In Example 2, the AI responds with a lengthy message on the full return policy, only half of which is relevant to the customer.
In some instances, it doesn’t make sense to ask questions and you don’t want to prolong the time it takes to deliver the information your customer is seeking, but in the right circumstances, this can add a nice conversational element to your experience.
3. Use Interjections in Your Conversational Design
Another way to make the experience less robotic is to acknowledge things that a customer says before moving on. Deploy interjections throughout your conversational design plan like “Thanks for the info”, “I see”, “Understood”, “Got it”, etc. can make the experience much more human-like.
4. Be Empathetic
An AI can be trained through conversational design to be empathetic, triggered by specific words, use cases, and customer sentiment. While an FAQ related to a package that is delayed might simply state how a person should proceed, an AI Agent could first offer up an apology acknowledging that the customer is likely frustrated. A simple “I’m sorry that you’re still waiting on your order. Let’s find out what’s going on…” can go a long way.
Identify the scenarios in which a person might be frustrated/unhappy and craft copy that acknowledges their feelings. This can also help you decide which instances should be better handled by a human agent. If you sell baby strollers, you can anticipate that a customer is going to be very upset if their child was hurt or in danger while riding in your product. Instead of providing refund or warranty information via the AI, elevate to a human right away and say something like…
“We’re so sorry that you experienced this. As safety is our top priority, I’m going to have one of our human agents assist you in this serious matter.”
5. Inject a Personality
Always stay true to your brand and tone of voice, and give your AI a personality. Train your AI to engage in small talk and be witty (at the right times), but never at the expense of providing the information that consumers seek quickly. Be prepared to do things like telling a Joke (you’d be surprised how often AI agents are asked to do this), solve a simple math problem or respond to an invitation to go on a date. Have fun with it, but always bring it back to the core purpose.
“I’m not allowed to date until I’m 16. I can, however, assist you with any of these issues…”
Related to personality is naming your AI. Some brands opt to give their AI a name like Bank of America’s Erica or WestJet’s Juliet, while others opt to refer to an AI as a virtual assistant. We’ve seen both approaches work brilliantly. We highly recommend, however, never trying to pass an AI Agent off as a human agent.
6. Think Beyond Text in Your Conversational Design
If it aligns with your brand, think about sprinkling emojis, images, GIFs and other content into the conversation. Emojis are today’s digital language and can make a nice addition to a response. See here:
- Your order is on its way.
- Your order is on its way 🙌
7. Use Contractions and Minimalistic Punctuation
The conversational interface provides the opportunity to be more casual. AI Agents should use contractions like “Let’s see what I can do” or “Don’t worry. We’ll figure this out together.” This is much easier and more natural to read in a chat.
Furthermore, don’t go overboard on punctuation; it can be used minimally. If you’re answering a question in a few precise words, you don’t necessarily need to add a period in the end. Also, limit the number of exclamations your agent uses. (Just imagine getting this message – Happy to help!! Let me know how else I can assist you today!!)
8. Be Upfront with Capabilities
While AI is becoming more prevalent in customer service, many of your customers might be engaging with a virtual agent for the first time. Briefly educate them on the best way to ask questions (be concise). Additionally, make sure to let customers know how an AI can help as well as how they can get in touch with a human. Transparency is critical to minimize frustration.
Creating conversational copy for AI-powered customer support requires a different approach from writing FAQ or agent training manuals, as well as from creative marketing content. Don’t get caught up in trying to be too creative at the expense of not resolving tickets quickly, but also understand that conversational AI is a new interface that should be engaging. Have fun with it!
Thinking about launching an AI-powered customer service? Let us help. We have copywriters on staff to help you design interactions that are both meaningful and engaging.