High expectations
Expectations of chatbot capability are high. We expect that they will understand us and have the answers. When they don’t, frustrations surface.
When a chatbot fails to understand, gives responses that sound scripted, unnatural, inaccurate or irrelevant, is unable to detect tone, takes too long to reply, or too many steps are needed to get to the answer, the user is dissatisfied and frustration creeps in.
These are just some of the frustrations, amongst many, that surface from bad chatbot experiences. The business impact of these frustrations include harm to reputation, reduced engagement and decline in trust.
It’s not as simple as “Just add a chatbot”
Organizations expect the much talked about benefits of a chatbot to fully materialize on day one. Whether the chatbot is being deployed for better customer experiences, greater efficiencies, improved productivity, or increased lead generation and qualification, the launch is just the beginning. Organizations and executives who think it is a one-off project will be disappointed.
Chatbots need to be kept relevant, accurate and helpful. This requires a proactive maintenance strategy built-in to the design of the chatbot. With a deliberate anti-aging maintenance strategy, long-term continuous business value can be achieved.
Conversational AI systems require more than just technical maintenance. They deal in language, emotion, and knowledge and each of these change and evolve constantly, and user expectations are continuously shifting. So too, therefore, must the chatbot, voicebot or virtual assistant need to evolve, change and shift, to be of value both to a business and the user. This requires maintenance, upkeep and attention.
Proactive, anti-aging maintenance
There are four key areas where continuous anti-aging maintenance is required to avoid a chatbot quickly becoming obsolete and causing user frustration and reputational damage.
Anti-aging maintenance by design must address:
- Language
Language evolves extremely quickly. New words emerge, slang develops, tone appropriateness shifts, culture and humor evolve, and even punctuation changes. A chatbot speaking ‘2020’ language will feel outdated to the user and will be less trustworthy. When a chatbot sounds out of date, trust is eroded, even if the information is factually correct.
Yes, chatbot knowledge and accuracy is extremely important. But to be relevant it must communicate more than just correct information. Chatbots must also match the tone of the user, understand context and subtext, and sound ‘conversational’ and not robotic. By ensuring language is constantly being updated will help avoid creepy and unnatural interactions.
Chatbots will age socially unless continuous attention is given to language maintenance. Language updating is not a cosmetic exercise. It sits at the core of trust, usability and relevance. Linguistic aging can be mitigated by designing for conversational AI anti-aging from day one.
- Product Content
Over its lifespan a product changes in numerous ways. For example, new features being added or functionality removed, new plans or current plans renamed, or use and terms policies updated. These are just some of the content related issues that can result in conversational misunderstandings, the sharing of wrong answers, or even workflow hallucinations.
When products and content are updated, this information needs to seamlessly be available to the chatbot replacing outdated and obsolete information. Achieving this as a seamless process requires methodically reviewing content at regular intervals, supported by a proven content update pipeline tool.
- Human-in-the-loop
Conversations are not always straightforward or take one direction. Every conversation is different and by incorporating human-in-the-loop for complex queries, unexpected user inputs, and ‘tricky cases’ accuracy and brand consistency will be maintained.
- Recurring maintenance cycle
The simple idea to ‘just add a bot’ that sounded like a great idea will quickly become a nightmare, unless a regular maintenance cycle to review, test, update, refine and evolve the bot takes place.
Using metrics that track user experience, and include misunderstandings, abandonment and sentiment and undertaking regular performance audits to catch signs of aging. Bots can be updated accordingly and will continuously evolve as users and products evolve. Implementing a maintenance cycle ensures chatbot relevancy, accuracy and reliability.
Chatbot maintenance must go beyond tech maintenance. A chatbot should be a long-term strategic asset and needs to be looked after as such. The value of a chatbot is inextricably linked to the time and effort committed to social, cultural, content and language anti-aging maintenance.
Chatbots that are left to ‘age’ slowly but surely lose trust. Brand engagement will reduce and reputation will be damaged.
Chatbots need to be constantly tuned to the times; staying socially and culturally on point, armed with the most recent knowledge, and speaking the language of the user.
Embedding the practices mentioned above will make your chatbot more than a feature or novelty. Keeping your chatbot young and relevant requires maintenance by design, and ensures it is a value-adding organizational asset.
