13 Oct Guest Q & A: How does Conversational AI Impact Contact Center Operations : Part 1
In this Q&A guest-post, a well renowned Principal Analyst@Forrester Research serving application development and delivery professionals contributed his thoughts on the trends, pitfalls and best practices of AI-chatbots on customer experience(CX) and customer support operations. This blog is part 1 of the two-part Q&A series.
1) What’s your opinion on the AI chatbot trends in customer experience?
Firstly, there is a big difference in what is going on in AI and customer service vs. what is going on with chatbots in customer service. For more details on what is happening with the broader set of AI tools, please see Kate Leggett’s report “How AI Will Transform Customer Service.”
On the chatbot side, brands continue to look to this technology as a cost-savings play, at least in customer service. While we believe this is a wrong-headed way to approach chatbots, this attitude has some major impacts on how vendors must operate. They need to increasingly focus pricing models on the effectiveness of deflection, for example, and less on pure interaction volume. Or at least, they need to create all-in pricing models.
Brands are still often running pilots with multiple technologies and many have still yet to decide whether they want to use a development framework (with either internal IT resources or outside partners doing the actual bot development) or a somewhat more vendor-focused/more packaged approach. There is likely not going to be a single winner in this race. And when Microsoft now offers a packaged chatbot as well as a development framework and component technologies, the picture can be cloudy even if a brand chooses a specific partner.
We have seen some good results for early stage projects, but no real slam-dunk case studies that show chatbots replacing the need for human agents. But for the basic FAQ-type questions, the simple rules-based chatbots can do a decent job. But for anything more complicated, brands end up spending a great deal of money developing bots capable to discerning customer intent. Often the difference in cost between these two types of deployments can be a factor of 5 or even a full order of magnitude.
Finally, we are starting to see some brands look for ways to use chatbots as tools to augment, not to replace, agents. KLM is one of the most widely-publicized examples. But as chatbots and other self-service tools take the simple questions out of the funnel, agents are left to deal with only the more complex and high-value issues. So, it just makes sense to provide them tools that can help bolster their performance in such a situation.
2) Which industries would be most affected by conversational AI trends ?
There are a few ways to look at this. Firstly, any industry with a lot of repetitive questions or processes can obviously benefit. This, however, is not really different from answering the question what industries could benefit from a speech-enabled IVR. Password resets? Yup. Shipping status? Yup. Flight or trip status? Yup. All of those can be done via IVR and all can be done even better with a conversational chatbot.
Secondly, there are some specific industries that see the biggest potential here. Insurance companies are all poking and prodding at chatbot technologies seeing if they can use the technology to answer basic FAQs, but also to help customers fill out forms and applications, and even more ambitious use cases such as doing the entire application process end-to-end. One auto insurer, for example, is running a pilot in a single US state to see how well a chatbot can handle getting an insurance quote done. This involves not just form filling, but also determining a prospective customer’s risk tolerance and checking for all possible discounts. Beyond insurance, financial services (which would include the financing arms of auto manufacturers, and other credit providing organizations), telcos, and cable/media companies are all looking at ways to use chatbots for selling additional products and for allowing customers to check on eligibility. In the telco world, for example, a chatbot could check your eligibility for an upgrade to your mobile phone and then help you through the purchase process for the new phone. Cable companies are looking to have customers check for a specific program or channel, and if the customer does not have that channel in their package, use the chatbot to sell them access to it (and to initiate the process on the back-end to turn on that service).
Finally, there are a lot of technology companies looking at tech support use cases. Today, these are often at the triage and route to the appropriate agent level, but the idea is to quickly move to providing the basic levels of support through the chatbot.
3) What are some of the uncommon use cases related to optimizing contact center operations using conversational AI. Can it be industry-specific ones as well?
Some forward-thinking companies are looking at models beyond just basic cost savings. For example, one cable company has deployed a chatbot to handle simple interactions and reduce the volume going to their outsourced and offshore chat agents.
But instead of just taking the savings and improving the company’s P&L, it is taking that extra money and using it to fund a program to move those agents from offshore to onshore. So, the chatbot is the mechanism to improve the customer service experience for customers with more complicated issues that require a human being.
Another example: An insurer has both licensed and unlicensed agents. The unlicensed agents, by regulation, cannot handle all types of customer inquiries. Those inquiries go to the (much more expensive) licensed agents. But, after some analysis of trends, the insurer came to believe that the unlicensed agents were escalating far too many customer inquiries. So, they developed a chatbot that first helps the unlicensed agents determine whether they can handle the interaction. When they can, the chatbot then helps support the agent (either through dialogue, sending the process to the agent, or displaying a PDF). When the unlicensed agent really cannot handle the interaction, the chatbot routes the interaction to the best available licensed agent.