Part 2, Chatbots today: The Three Primary Use Cases
This is Part 2 of the three-part Executive’s Guide to Chatbots in 2018. To view Part 1, click here.
Despite the disappointments of 2016, chatbots have now been integrated into key areas of the customer journey to expedite processes and offer intuitive UX experiences. Unlike Tay and Eliza, the best enterprise chatbots have narrow applications and specific purposes. These are the three areas in which chatbots are currently offering enterprise value:
1. Customer Service
Last year, NewtonX collected insights from 100 top executives across multiple industries. Each company generated over $1b in revenue and received millions of customer service inquiries per year. Because chatbots have been lauded as the solution to scalable customer service, we took a look at each company’s spend and ticket volume across channels across channels, and also gathered qualitative opinions from top executives on how well this spend has been paying off.
The study found that at the time, chat was still an expensive customer service medium and accounted for only 11% of customer service volume split. Still, executives cited chat as the No. 1 CSS trend for 2018, and indeed, Zendesk and startup competitors such as Helpshift have rolled out custom and pre-built chatbots.
These chatbots are designed to improve time to first response, scalability, and to automate the rote work that agents perform. For instance, much of customer service agents’ time is consumed with collecting basic information such as order number, name, shipping address, etc. Bots can collect this information in a fraction of the time and either use it to answer the customer’s question or can synthesize the information into a snapshot for an agent.
2. Sign up/ onboarding
Many companies use chatbots to replace basic webforms on mobile, tablet, and sometimes even web. Like the information collection chatbot described above, onboarding chatbots collect sign-up information in a dynamic and user-friendly way. Sometimes, they employ buttons for binary options so that users do not have to type in responses and can instead just tap the option that best describes them. For instance, Lemonaid Health uses a dynamic onboarding system that combines forms for information like zip code, and buttons for information like gender.
These chatbots use simple decision trees and usually do not use any form of NLP or machine learning. In other words, they are not “smart,” but still offer a faster, simpler process for sign up and onboarding, particularly on mobile.
3. Dynamic search
Airlines and hotels employ chatbots to help users find flights and rooms without having to navigate and scroll through mobile-unfriendly options. Through posing a series of questions and answers, the bot can surface the best flight for the customer based on parameters and preferences. Icelandair, KLM, and Expedia.com are among the numerous airline providers to offer chatbots in lieu of traditional search.
The e-commerce industry has also experimented with using chatbots for search, but the attempts have been somewhat lackluster. eBay, for instance, launched its ShopBot in 2016 to allow users to search for products by describing their needs to a chatbot, but ended up shutting the bot down in September of 2018.
Dumb but functional: what these use cases have in common
The reason that chatbots have been successful in these three enterprise areas is that they have a narrow purpose with limited potential avenues for the “conversation” to go awry. In 2016, chatbots were too smart; today, chatbots have gotten dumber but more functional.
In part 3 of this series on chatbots we explore the future of enterprise bots, and what executives need to do to prepare for this future. The insights in part three were gathered from top executives who have invested in, written about, or built enterprise-ready chatbots.