Before creating the dialogues and conversation flows with your AI agent, it’s important to take some time to decide on your AI agent's tone of voice, and match these with your company’s identity. It’s also good to become familiar with some important conversational elements that work well in chat. Your AI agent is likely to be the first point of contact when your customers have an inquiry or need help, so it’s crucial to think about what kind of an impression you want to leave.
Before setting up dialogue flows and hypothetical questions, spend some time thinking about your future AI agent’s tone of voice - how is your AI agent going to fit into the overall brand and image of your organization?
Important - At the end of the page, we have an exercise that you can download to work with your team to build your replies with some of our recommendations.
Be Consistent as a Company
An AI agent, just like your customer support team, is an extension of your brand. Therefore, it is important to match your AI agent with your company’s brand and identity.
What’s the voice of your brand and how do your customer support agents communicate with your customers?
Consider your AI agent as an addition to your customer support team and match its voice to your brand and agents’ voices to maintain consistency throughout all customer interactions. If your brand is traditional and more on the formal side, you might decide on a polite and formal AI agent. However, if your brand is targeted more toward a younger audience, going with an energetic and more informal AI agent might be more suitable.
After thinking about these two points, you should have a better idea of what kind of tone of voice suits your AI agent best.
Have a look at the examples we created for you here.
Rather than attempting to invent an entirely new mode of communication, why not build off of what you already have? Some organizations have standardized communication guidelines for their customer care agents, and these can be a great reference when first setting up your AI agent. Should you not have any such documentation, though, a combination of speaking to your agents and reading past tickets could also work well as a starting point!
The voice is not a simple thing, though. Many elements combine to make up the outward face of your organization, and consistency is king when it comes to giving a good and professional impression. Here’s a small, non-exhaustive list of elements that you may want to consider:
- Punctuation practices
- Capitalization
- Spelling practices (i.e. American vs British English)
- Standard greetings and goodbyes
- Abbreviations and acronyms
- Any special vocabulary that you use
A good example of where your voice might factor in is your immediate response when a customer complains or requests a refund/similar. Do you apologize, or do you jump straight into trying to find a solution? Whichever approach your agents use, that’s most likely what you’ll be wanting to bring to your AI agent.
Another example is the extent to which you refer to the customer’s name: some customer care departments have set a golden rule of 3 namedrops per conversation, while others aim for even more, or perhaps only one! How many will your AI agent be making?
However, there are still some points on which AI agents differ from real agents, one of being that the AI agent will likely lack the same nuance as a real agent. For that reason, we recommend that your AI agent be assertive when collecting information, and empathetic whenever a customer has had a poor experience.
Be Human Centric
AI agents are designed for humans, by humans. Understanding who your customers are will help you better design your AI agent.
Is your customer base mostly made up of the Millennials (born 1980-1994) and Gen Z (born 1995-2010) who are quick learners and appreciate self-help? Or are your customers made up mostly of Generation X (born 1960-1979) and the Baby Boomers (born 1940-1959) who are likely to appreciate an intuitive AI agent with easy access to speak to a human agent? Or perhaps your customer base is a mixture of demographics, in which case you should consider designing a solution that is suitable for all.
No one knows your customers better than you - so use that information to your advantage. What do they like, and what do they want to gain from their interaction with your AI agent?
Marketers often like to refer to the “Generations”, and it is important to relate your AI agent to the age of your customers, too. Here, we’ve written out some recommendations for the latest four generations.
Baby Boomers |
Generation X |
Millennials |
Generation Z |
1940-1959 |
1960-1979 |
1980-1994 |
1995-2010 |
Select an AI agent that can also handle email communication |
Ensure that the AI agent platform is intuitive |
Offer an automated FAQ or knowledge database |
Ensure your AI agent platform is mobile friendly |
Train your AI agent to provide an easy path to speak to a human agent |
Make your AI agent offer incentives in its customer communications |
Write your responses to make Millennials feel like part of your brand’s tribe |
Make sure that your AI agent is set up to answer queries 24/7
|
Write your dialogue to be less colloquial and more formal |
Offer incentives to direct the customer to the live chat. For example, you could tell them how long the waiting time will be before they can speak to an agent on the phone |
Personalize your AI agent's dialogue, making it short, snappy, and conversational |
Provide AI agent responses that give the user agency to find their own answers |
However, this may not be the same for every culture, these definitions may not apply in entirely as clear-cut a fashion as we put them. Do a little bit of research, and you should quickly be able to have a similar plan of your own.
Lastly, it’s important that your current voice meets your customers on your customers’ terms. Do they prefer a more formal or informal approach? A more friendly and warm dialogue, or one that’s more pragmatic and straight to the point? While real agents can easily gauge clients and speak to them in a way that they understand and appreciate, your new AI agent doesn’t have that ability. Instead, we have to create a AI agent that’ll be able to communicate with all of your customers.
Our Recommendations
Regardless of the formality, tone, and language of your AI agent, there are some general guidelines and recommendations that we recommend you follow.
Be honest. Establish the right expectations right from the get-go.
Your customer is not speaking to a human agent, but rather an AI agent. 70% of American customers prefer speaking to a human agent rather than an AI agent, and the last thing you want to do is lie to them. Tell your customer that they’re speaking to an AI agent - and make sure that they can easily escalate to a human agent.
It’s also important to be transparent about the fact that the AI agent can’t do everything your customer may need, but rather a subset of tasks, for which reason we recommend that you’re direct and confrontational as to the AI agent's capabilities.
Be direct.
Make it clear what the AI agent can and cannot do.
Good ways of demonstrating this can be through the use of buttons or by listing the topics that the AI agent can handle.
Make sure your AI agent doesn’t overpromise. The confidence that a human agent may have when handling a given situation cannot be emulated by the AI agent, and the last thing we want is an unfulfilled promise. For example, instead of saying that something won’t take long, let the customer know what the actual set timeframe is - even if that’s a lot longer than they might want to hear.
Be concise.
Your AI agent's responses have to be general, direct, and accurate, all at once.
The AI agent's response has to be general enough that it can be seen as an acceptable reply for all the possible messages within the intent, while still being short enough that it’s readable. If you have an FAQ, then most likely you’ll have the majority of the information you need to fill out your AI agent's conversations. However, pasting your FAQ answers directly into your dialogue flow is hardly a good idea. Instead, increase the ease of reading by rewriting the content into a more concise form.
Be clear. Stick to one thought at a time and separate points with line breaks. Large chunks of texts can be overwhelming, and even more so if not separated by points. Dividing a large message into multiple small ones, and using line breaks, increases readability for your customers, and also makes editing your dialogue easier in the future. Similarly, you ideally want to avoid using large, long, and complex sentences. While short, simple sentences may appear less aesthetically appealing or eloquent, they are much easier to quickly skim through while looking for information than their longer counterparts.
0 comments