Strategy 101: Explaining the three facets of chat
We’ve all run into bad chatbots. It’s the bot that seems to only ever respond with “I’m sorry, l didn’t understand you”, or an assistant that’s so useless you’d be more productive working without it.
While it may feel like bad chat experiences are the rule, the technology and expertise are available to create a chat channel that is truly valuable for both businesses and their customers.
So what really makes a good chat experience? Let’s have a look at the three essential strategy pillar of an exceptional bot:
- AI (artificial intelligence).
- SI (systems integration).
- UI (user interface or conversation design).
1. Chatbots need intelligence (even the artificial kind)
The first essential component of a good chatbot is artificial intelligence capability. With the explosion of the market jargon around AI, it can be difficult to understand exactly what it refers to. So what specifically is AI and how does it relate to chatbots?
AI can be broadly defined as an algorithm that has the ability to perform complex tasks, typically associated with human minds. This could include:
- Problem-solving.
- Learning.
- Perceiving/interacting with the environment.
Within chatbots, AI capabilities are primarily focused around natural language processing (NLP). This is the analysis and comprehension of human language by a machine and can include both speech and text. Intelligent neural networks work to predict the intent of providing statements or queries, extract entities from these questions and then generate appropriate responses.
For example, the queries: ‘where is the nearest pizza place?’, ‘find me pizza’ and ‘pizza near me’ all have the same intent despite being totally different sentence constructions. NLP allows chatbots to derive this intent and formulate the best response – in this case by giving a list of nearby pizza joints.
Chatbots without this AI capability are unable to ‘understand’ language in any way. Instead, they rely on fitting themselves to preformatted flow diagram responses. This basically means that they will be unable to answer, or even function, with anything outside a rigid scripted set of questions followed by a set of responses. If a customer doesn’t hit certain keywords or even exact sentence construction in their query, the chatbot will be unable to fo anything.
The difference here in the customer experience is clear: Chatbots working without AI are not only frustrating for customers, they ultimately fail to fulfil their expectations. Customers expect chatbots to understand everything they say, and if this is not the case the resulting experience is beyond a disappointment. Chatbots without AI are almost guaranteed to erode your brand perception and prevent you from achieving the trust value that this delivery channel has to offer.
2. To deliver value you need to integrate your systems
The second pillar of chatbots is successful systems integration. By this, we mean that the chatbot must be able to seamlessly link your existing systems in order to provide value for the customers interacting with it.
Why is this essential? Because ultimately chatbots work as a part of your value delivery. While they’re a new and exciting channel, they must be able to provide customers with access to the service which your business performs. This could be selecting and buying tickets or connecting with products or advice. Whatever your offering, your chatbot needs to be integrated with it.
Imagine how frustrating it would be to interact with a travel chatbot, hoping to find some food flights to New Zealand, only to find that the bot can’t access any key information (such as listed flight times and prices) or book them.
Chatbots are not an entity on their own, they’re a dynamic channel to extend the delivery of your service. To do this successfully there needs to be seamless integration with your value delivery systems.
3. Good conversations are designed
Finally, good conversational design and UX are the missing pieces in many chabot experiences. Conversational design can be best described as a synthesis of disciplines such as script-writing, visual design, audio design, interaction design and UX design to create and curate a conversational flow. Conversational UX works to implement an enjoyable, flexible conversation that enables a great overall experience for the user. It aims to make the chatbot behave in a conversational way rather than merely as a transactional platform with a few tasks in mind.
Conversational design is particularly essential in tandem with natural language processing. NLP is able to discern the intent of the customer’s question and then place them within the correct conversational flow as designed by the conversational designers. This allows for a more responsive conversation which can bring personality and voice to the chatbot’s responses, and also include useful visual elements in the form of widgets or carousels. It’s particularly useful for actions such as selecting a seat where there is a compulsory visual element that must be interleaved with the text interactions. Such a design makes the user experience enjoyable and seamless.
Chatbots that lack conversational design are purely transactional and provide no form of experience. Responses will be very mechanical sounding and will not convey any brand personality or an identifiable voice. Also, without AUX features such as carousels and extension widgets, much of the burden is placed on the customer to spell out exactly what they need and want.
Why do we need all three? An example
All of our discussed pillars are essential to create a fully functioning chatbot ready to delight and deliver value to your customers. To demonstrate the collaborative nature of thee elements let’s check out an example:
A drink retail chain has a chatbot accessible via Facebook messenger. It allows customers to browse their product selection, ask for information about the deals and products they want, find their local store or even order their usual beverage to their doorstep.
- Customer types in question ‘do you have any Shiraz from Barossa Valley on sale for under $30?’
- The AI pillar determines the intent of the statement (to find wine products catalogued currently as on sale, with a varietal type of Shiraz and location set to Barossa Valley).
- The AI pillar then places the customer in the product conversational flow and pre-populates data fields with all the information it has extracted.
- Meanwhile, the systems integration finds the required products and delivers them to the chat platform.
- The conversational flow will then present the information to the customer in a creatively designed format (a list of the products in a horizontally scrollable carousel).
As the conversation develops, this process will repeat in a variety of ways in order to create a seamless conversational experience. Removing any one of these pillars will result in a failed chat process.