Bots are gaining popularity because they are effective in providing good customer experience. As bots never sleep you always have the door open for your customers i.e. 24×7, 365 days. Although businesses love bots, but often customers hate them due to legitimate reasons. The most common mistake bots do is that, they pretend to act like humans, and they miserably fail on it. Bots are not a replacement for humans irrespective of how good they are on natural language understanding and speaking. Bots help to provide a communication channel that is more available and more efficient for few well defined tasks.
As it is quite easy to create a bot these days, business jump to create one of their own. But most of them lack purpose and effectiveness. I would like to capture few important qualities of an highly effective bot.
Clear bot identity
A bot should declare itself as a bot. This really helps the bot because expectations levels are set upfront and limitations are understood. On the other hand, if customers realise during a conversation that they are talking to a machine and not to a human, it can have a negative impact. A bot can and in fact should interact like humans but making its identity clear helps both the sides.
A bot can and should have a persona. It is easy to talk to someone whom you understand and when you know the personality of the other party. Depending on the purpose, the persona can be adapted. A good bot persona consists of interactiveness, respectfulness, trustworthiness, consistency, etc. Also adding personalisation to the persona is a good way to engage the audience.
Natural Language Understanding
People think that bots are a new thing but bots have existed since the beginning of computers. Traditionally, all interactive computer programs were bots but they didn’t understood human language. And they defined their own communication rules/pattern , for example : enter ‘1’ to do this, enter ‘x’ to do that, etc. But with Natural Language Understanding(NLU) technology maturity and availability, this has given a new life and meaning to bots. NLU is essential these days for bots that engage humans in a conversation. NLU training should be a continuous exercise capture user intents precisely and cosistently. As languages are vast and deep, it is a daunting task to learn languages for both humans and for bots. That brings us to our next important attribute of highly effective bots i.e. purpose
A bot is not a human and lot of humans are also not good at effective communication. The main reason for that is lack of purpose in communication. If someone communicates without a purpose then it is hard make any sense. If we define a clear purpose then developing a bot is easy, effective and meaningful. What do I mean by that is, not all bots are suppose to tell you weather info, or to tell you a joke. There are some general purpose bots to help us with that information. Generally, bots should have a clear purpose such as customer service, help to buy/sell, help a person in need, etc. Once we define a clear purpose for a bot, then we can train it using the natural language. More and more NLU training helps to capture the intent precisely.
Educate users and prompt with options
Bots often go into a completely non understanding mode. Often bots frustrate us with ‘not able to understand’ messages. No matter what you try to say, you keep getting this response from the bot. This can happen because of variety of reasons such as like lack of clarity, purpose, wrong identity, poor natural language understanding, etc. Sometimes, as a bot developer/provider you think that you have covered the above mentioned attributes well, but users keep surprising us. The reason is the gap between bot functionality and expectations. To bridge this gap, it is essential for bots to educate their users. It is good to provide upfront introduction about bot capabilities with a welcome message. Educate bot users what they should ask and what they should not, or simply explain what this bot is capable to do. Providing options to users with prompts is a good trick to drive the conversation properly. As users get more familiar with the bot, it is easier for them to communicate with it.
Continuous learning and personalisation
Users find it exciting when bot remembers them and remembers what they generally need. Technologies like machine learning can help bots to learn the user behaviour and to provide correct predictions. This can help bots to drive the conversations better and increase the customer engagement. If both bot and user are able to learn about each other and can start to predict correctly about each other well, then the conversations will be more natural and purposeful. Learning makes the personalisation easier and personalisation takes away the robot like communications. Bots should not claim to be humans but should not act like boring robots too, right balance between the two is what makes a great bot.
Reasonable response time and interim responses
We humans are impatient and somehow we are more impatient when we talk to a machine. So it is important for bots to respond in a reasonable time. If there is something that bot does behind the scene and may take longer time to respond then it is good to let the user know about it. For example a bot can say “Let me get that, but it may take couple of minutes, so please standby”. Also if something is talking longer than expected, then interim responses can help to keep the user engaged.
Use of alternate channels wherever necessary
Understanding your bots limitations is more than useful. Once we know the limitations, we can do something about them. Using alternate channels is a good way to provide your users what they are looking for when the bots cannot. If different channels can work together and contextual handoffs are possible then customers will always be engaged and you will be able to drive them purposefully.
Short, sweet and refreshing content
A good speech is the one that is short and sweet. Bots should converse clearly and concisely but in an interesting and refreshing manner. Multiple random phrases for a purpose help to keep it to be non-boring. It is essential to refresh the content regularly but without changing the communication structure of your bot.
It is easy to create bots but it is difficult to create effective bots. An effective bot should have clear identity and purpose. Natural language understanding is at the core of bots and human like interactions are essential. Continuous learning and personalised content provides good customer experience and more fulfilled user journeys. Knowing the limitations and overcoming them through interim responses and alternate channels can help the business to keep the customers engaged. This article was meant to capture only a few of the qualities of highly effective bots. All of them may not be applicable for your bot and not all the qualities of bots are not mentioned here. So take what makes sense and build great bots. Good luck!