Despite so many benefits of chatbots, they are challenging to turn into success stories. We had explained the 3 reasons why your chatbot will fail: Discoverability, customer expectations and technology limitations. These are all hard problems but there are strategies you can follow to manage them:
Discoverability: Your chatbot can be on your website, on Kik, on Facebook messenger or on a myriad of other platforms. Until a dominant platform emerges like it did in China, you need to promote your bot as clearly as possible so your customers know where to contact you. Even better, integrate your bot into your customer service processes so it doesn’t even need to be discovered. A chatbot listening into your customer service calls, learning from daily interactions and suggesting your employees what to tell to difficult customers is not a dream. Numerous vendors like True AI or Wise.io already provide this service. So instead of creating new channels, you can augment your customer reps with chatbots and get some of chatbot benefits.
Expectations: Customer expectations need to be managed. I would avoid hyping up a new chatbot. Especially since chatbot capabilities are still relatively limited and since there are already websites full of screenshots of stupid chatbot replies.
Technological limitations: An AI interface that can solve all of customers’ problems without human interventions does not yet exist for any business domain. An AI system should either have known limits within its domain or human-like capabilities in that domain which leaves companies with 2 distinct choices:
- If you represent a small business or an enterprise for which live agents are not economically viable, brutally limit the domain of your customer service chat bot. Your customers will be essentially using a glorified FAQ service. If expectations are managed, this can be a fast solution to their urgent problems. And you can go a bit beyond glorified FAQ service with some very economical, self-service chatbot solutions.
- If you represent an enterprise willing to invest in chatbots and customer service, then enable seamless transfer to a human when needed. This approach is not scalable for M which is a learning experiment for Facebook but for a company already offering human customer service, this approach has significant cost-saving potential. You should probably talk to some vendors before deciding whether to inhouse the development or use a vendor. Given the boom in number of chatbot vendors, you may want to check out the list of leading chatbot vendors that provide end-to-end solutions.
How can we do better?
Your feedback is valuable. We will do our best to improve our work based on it.