ICMI is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


Five Tips to Successfully Integrate Chatbots in Customer Service


I think the first time I spoke about the future of AI and chatbots was the ICMI conference in 2015. The slide read, “The future as we see it: Chat, social media, texting, and in-app service will become the channels of choice.”

Since then, I have had conversations over the past five years with contact center leaders who said they didn’t think their customers would ever accept talking to a robot; they also weren’t sure the technology would allow them to provide an effective customer experience.

Many of those doubts evaporated with COVID-19. Choosing to engage their customers with chatbots and digital assistants was no longer a choice but a requirement. And the reason they took the leap to fully implement AI and chatbots was that the technology was ready – so, the success of pushing tactical calls to a bot – just made sense.

If, like me, you have been online in the last several months seeking to chat with a live person about an invoice or problem, you have probably experienced both good bots and bad bots. That is because, while the technology is ready for primetime, it isn’t being implemented in the best way possible. Companies are sometimes expecting the technology to do the work – and that is just not how AI and chatbots can best be used to deliver a smooth customer experience.

If you plan on implementing AI or a chatbot, here are five things to consider for building a good chatbot experience:

Be realistic about the scope of the bot.

You have tons of interaction data in the contact centers – so use that data. We recently worked with one customer who started with 10 interaction-types but quickly realized that two of those were much too complicated for the AI interaction. It’s necessary to be agile and iterative in your approach.

Choose the channel that fits your customer base – and choose the technology that provides the best platform.

Starting small and growing is better than trying to tackle the big stuff.

  • Salesforce has a strong play with its Einstein platform, which brings to life chatbots, routing, and even case gamification, inside their CRM functionality. The options are endless – as we recently implemented Einstein with the ability to chat in 14 languages.
  • AWS offers a strong AI platform with Lex (which is powered by the same deep leading technologies of Alexa). And, with their AWS Connect product, they can provide multiple ways to engage customers using bots – both within digital and voice channels.
  • IBM’s Watson connects seamlessly to most applications and provides a bigger platform that drives other parts of the AI experience – in addition to chatbots.
  • Most cloud telephony vendors also have their own AI-assisted bots. It may take some time to choose the right one for your current and future needs. Chatbots are more than just digital, as AI-assisted IVRs become better at understanding human voice.

    Planning is the most important part of the process.

    Designate a significant amount of time to understand what you are really expecting to solve - it’s easier than tweaking the technology after the fact to solve the problem. How will the bot relate to the user? What is the role and job description of the bot? How proactive do you want the bot to be? What data will the bot use to learn and engage

    Train the chatbot well – and then train it again – and again.

    Think of a chatbot as a new employee who you send to four to six weeks of classroom training – and then follow-up daily with huddles, conversations, and on-going training. The bot will only be as good as you teach it to be.

    Now that you have a bot ready, be sure to train your agents on how it works, what it knows, and how their interactions may change based on the implementation.

    Just like when we created FAQs on websites in the late ’90s and moved content to a self-service portal in the mid-2000s, chatbots will change their interaction paradigm. Agents may be asked to pick up on step five of the seven-step process that the bot provided. Ensuring your agents understand the way the bot thinks and engages with customers will make both them and your customers less frustrated.

    All of these steps are just the beginning of the chatbot path. Whether you are in the retail customer service world with 50 calls on hold and 45 chats in queue, or are in a business that is just coming back to life after the initial reality of COVID-19, now is probably the time to look at how AI and chatbots can help your customers and your agents.