Understandably, many customer service leaders dipping their toe into AI for customer service are reluctant to admit it. After all, the stats still suggest that most people would rather speak to a person than an AI.
On top of that, people have the mistaken belief that their agents would appreciate AI. The truth there is that, if brands take a balanced approach to humans and AI as DigitalGenius believes, then they can achieve harmony. After all, our own research says that agents don’t like doing the repetitive work that AI is best suited to.
In the past, the quality of chatbots would mean that it would be pretty obvious when the response was from a bot rather than a person (unless the agent script was particularly robotic). But now, with GPTs and LLMs making AI conversation so natural, it is possible to disguise the fact you have an AI chatbot.
So, should you tell your customers you have an AI chatbot? Or should you disguise it with a name that looks like a real person? Ultimately, it’s a brand choice, but here are some arguments to consider.
Establishing trust with your customers
At the heart of a great customer experience is trust and so it’s important that your customers trust you. Therefore being open and honest with them should really be the best way forward.
There are many ways to do this. Air up have a particular brand tone of voice, and so their bot Airic has a particular way of introducing himself, which you can see below.
Another way is to just simply state that it’s a bot, as All Saints have.
Speed of interactions
If you admit that you are using a bot, then the responses can come as fast as the processing allows. If you are keen to hide that fact, then you may need to take steps to slow down the responses so it looks as though a human agent is processing your request and typing a reply.
Obviously the bot can then still be fast, but it’s an extra step that has to be taken if obscuring it is a factor.
If you choose to be up front, then the bot can be as fast as it can. You may still want to slow it down a touch so if you send multiple messages at once the customer can read them comfortably. But customers expect bots to be fast, so this is less of a concern.
Be prepared for your CSAT to be affected (positively and negatively)
CSAT (Customer Satisfaction) scores can be affected if customers know they are talking with a bot. Let’s imagine a situation where the bot is doing exactly what a customer service agent would do.
So, if a customer asks to return an item after the date for returns has passed, an agent would be correct to say that it wasn’t possible. In the same way, bots should not be making up company policy and allowing a little flexibility.
But if that annoys the customer, they are more likely to mark the bot negatively than the agent. Just as you would (hopefully) never hit your colleague for doing something wrong, you might lash out at the company printer that isn’t cooperating. The machines have to take the hit, and it’s the same in customer service.
Now, a chatbot should be able to answer lots of other questions very well and can actually increase CSAT by giving quick responses to common questions. Likewise, by clearing the backlog of simpler queries, the AI can allow agents to get to the complicated customer questions faster. So there is a counterbalancing effect.
By drilling into what queries generate what CSAT scores, you can then start to triage and route the queries to the best place. So if agents are significantly better at dealing with product faults, then you pass customers to agents at the first opportunity and bypass the bot completely.
The opportunity for brand tone of voice
The rise of generative AI chatbots has allowed brands to create bots that match their tone of voice. If you are a brand with a strong identity and trying to communicate that to your customers this is a great way to start that.
You can prompt your bot to be formal, or you can make it playful and use lots of emojis. Because of generative AI the actual responses can be different each time, allowing you to make the whole experience a strong part of the experience.
Conclusion: Yes, you should tell customers that you use AI in your customer service
On balance, the answer for most brands should be Yes. The statistics that suggest that customers dislike talking to AI more than agents are true but the reality is this: customers don’t hate chatbots, they hate bad chatbots.
The overall level of quality in chatbots is going up and up, and customers are getting used to them. Bad ones that are built to block you and deflect you are still prevalent, but are being usurped.
If you are going to use AI, you should be sure that it’s effective. That is, you use it to do the tasks its best placed to do, and you keep your human agents in the loop to do everything else. When you do that right, you’ll want to shout about your chatbot.
To find out how you can get a chatbot you want to shout about, talk to us today.
