How to Mature Conversational Experience Development

As new as building out conversational experiences are, tried and true software development practices still apply.

Today, conversational experiences such as voice assistants and chatbots are ubiquitous. According to a recent Modev report: “The market has accelerated from post-hype to a growth market…” The report states that funding raised by companies in this space in 2022 is expected to “significantly” surpass 2021 funding. (Conversational AI Supply Chain Report — Q1Q2 2022, Modev).

What Are Conversational Experiences?

Conversational experiences allow users to interact with digital systems using natural language through speaking or typing. The hands-free approach that voice assistants enable and the 24/7 availability of text-based chatbots are just a couple of advantages of conversational AI technology.

While companies may be bullish about conversational AI, users aren’t necessarily convinced: a recent US survey found that nearly 73% of participants would not use the company’s bot again after a bad experience with it. A few months ago, Christoph Boerner published an insightful article on CMSWire about people’s reluctance to use chatbots. He mentions that customers believe that bots aren’t as effective as companies claim they are, that customers want to be clear that they are converting with a bot, not a human, and that many customers believe that chatbots simply can’t be as effective as a human.

Despite user sentiment, conversational experiences are here to stay. Yet, for the sake of customer satisfaction and the future success of conversational experiences, it is critical that chatbot and voice assistant builders focus on the customer and their experience. Along those lines, certain myths need to be debunked:

Related Article: What’s the Impact of Conversational AI for Contact Centers?

Myth 1: Get an MVP Out ASAP and Then Iterate

An MVP (minimum viable product) refers to a first version of software that typically is released with limited functionality. It is certainly a good idea to start small.

Unfortunately, MVPs are often an excuse to release a product that is far from robust: the language model is not based on data collected in a responsible manner such as through sample collection or mining existing relevant sources, but instead the conversation designer may just come up with a handful of utterances. Compound that by not having a proper testing strategy in place, and the result is a conversational experience that users can’t really use and hence certainly don’t love.

Organizations may have the best intentions to review the live product’s analytics to itate and improve the product, but with over 70% of users stating that they won’t use the bot again after a bad experience, the future user base has just been severely cut down due to this approach.

A more desirable approach is to focus on the quality of the user’s experience before MVP launch, while still committing to improving the product based on data gathered from the live system. Focusing on quality pre-launch requires adhering to proper process, which leads us to myth number 2.

Myth 2: Conversational AI Can Avoid Tried and True Software Development Principles

It is true that conversational AI development is a little different, just like mobile app development differs from web development. Conversational AI calls for certain skill sets that are not typically needed in web or mobile development, such as conversation designers, AI trainers and data scientists. It requires product managers, developers and QA testers to understand how to define, build and deploy conversational AI applications.

Leave a Comment

%d bloggers like this: