When we talk about ‘chatbots’, we often conflate two different things: text-activated and voice-activated bots. It sounds trite to point out that they are, in fact, very different! Indeed, it’s rare to find a bot that can be activated by text and voice. So, given the cost of setting up a chatbot, which should you choose for your business to consumer (B2C) interactions?

Our SuperServe research for Avaya with 8,000 consumers in a range of countries found that most (albeit a slight majority) favoured voice over text when contacting an organisation. Compared to older consumers, those aged 18–34 were a little more likely to favour using text.
So which should you use? Well, the answer to that question depends on a mix of consumer preferences, the context, and the purpose of the interaction. The first step is to design a smooth customer journey for the consumers when using a chatbot – nothing cumbersome. Brands need to take real care over this stage as this is likely to be a deciding factor in the success of the chatbot. One piece of advice for businesses is to avoid the mistakes made with IVR systems over the past 20 years. Too many menu choices ended up frustrating savvy consumers: ‘IVR forests rather than IVR trees’.

The case for voice-activated chatbots
Voice-activated chatbots are ideal for when consumers are multi-tasking, such as when they’re driving, cooking or getting dressed. At times like this, chatbots can save people time. Voice-activated chatbots are particularly good for giving commands and instructions. They also work well when there is an element of search in their use.

Most people find it easier to voice their thoughts than to write them down. Apple tells us that Siri is being used more and more. When it comes to Smart Speakers, the consumers we’ve spoken to fall into three types. Some are using the technology enthusiastically. Others can’t find more than a novelty purpose. And a few are even suspicious (is it somehow compromising their personal privacy?).

The case for messaging-activated chatbots
Privacy is one of the biggest advantages of messaging-based chatbots. If you’re in a public place and don’t want to be overhead, they’re the best option. The response from the chatbot can also be much more detailed (e.g. referring you to a specific web page). Using a messaging-activated chatbot hand in hand with browsing a website or app seems a natural way for consumers to interact with businesses.

Some people also say it’s easier for AI and chatbot technology to understand the written word rather than a voice command, bearing in mind the range of accents and interference from background noise. But some believe the opposite is true – that it’s harder for AI and chatbot technology to understand the written word, citing poor handwriting!

Whatever consumers think, messaging-activated chatbots are delivering results. Lloyds Bank says that up to 30% of customer requests on its banking app are now being handled by its textbot. And at its recent F8 developer conference, Facebook announced that more than 40 million businesses are using Messenger (see our blog) to communicate with customers. Amazon and Apple also point to success with voice-activated services.

7 ideas for implementing voice or text-activated chatbots
So here are a few things to think about when deciding whether to go for a voice-activated or text-activated chatbot for your consumers.
• Keep requests simple – shorter commands or queries from consumers are better
• Let the chatbot access personal data – this can help in terms of bookings or looking up customer details)
• Protect privacy – be clear about when a chatbot is likely to be used and for what purpose
• Consider context – what else will consumers be doing when using the chatbot?
• Use declarations – California law will require chatbots to state that they are not human
• Think about depth of information needed – how much information will the chatbot have to return?
• Consider consumer preference – do you know what your customers are likely to prefer?

With decisions like this, knowing your consumers really helps. You could get lucky, and get instant results from your chatbot. But in our experience, success is most likely when decisions are based on careful research with consumers. This then allows you to test solutions that can be developed and tweaked over time.