Date Published May 1, 2018 - Last Updated 4 Years, 351 Days, 11 Hours, 10 Minutes ago
Everyone is talking about automation, it seems. In fact, in a recent survey of their readers by ITSM.tools, automation was the Number 1 “hot topic,” beating out AI (Artificial Intelligence) by more than six percentage points.
Of course, no one likes to talk about the amount of work that goes into automation. The more sophisticated and complex the system(s) undergoing automation, the more care and effort need to be put into it to make it all work right.
Chatbots are fairly simple, right?
There’s a classic job interview question used by some organizations to discover how candidates think: Would you please walk me through instructions on how to make a peanut butter sandwich?
The candidate will often respond with, “Get two pieces of bread…” at which point the interviewer asks, “Where is the bread? How do I get it?” The candidate responds, and the interviewer continues to ask for more details. Where is the peanut butter? Is there only one kind, or do we need to specify “chunky” or “creamy”? Do we need a plate to put the sandwich on? Do we need a butter knife to spread the peanut butter? And so on.
Let’s consider chatbots—automated responses to customer or end-user queries. We envision the capability to have a user-facing chatbot that can answer basic questions and serve up links to helpful information. That’s great, and can certainly free up analyst time for more value-added work. But where does the chatbot get the information? If the chatbot can’t find the answer based on the user’s question, it’s useless. If it serves up the wrong answer, it’s worse than useless.
We’ve all seen impressive chatbot demos that cover just about every kind of question and response, but almost all of those demos are based on publicly searchable information, like the results of an internet search or on canned responses. What about your knowledge? Is it behind your firewall? Is it searchable by anyone on the planet? Chances are your knowledge is not accessible from outside your organization. So again: Where does the chatbot get its information about your organization and your systems? It comes from your knowledge repository and will only be as good as your knowledge. To provide information to automated systems, your knowledge must be accurate, complete, and up-to-date.
To provide information to automated systems, your knowledge must be accurate, complete, and up-to-date.
Your knowledge, by the way, does not consist only of the articles in your knowledge base; it can contain configuration information, directory information, asset information, and so on. And yes, it can contain information from internet search, as long as that information is accurate for your organization. In fact, the more data (pertinent data, that is) you make available to your chatbot, the better off you will be—as long as it is (say it with me) accurate, complete, and up-to-date.
The chatbot in our example is capable of providing answers much faster than a human can either type or copy and paste, but providing incorrect information faster is quite obviously not the goal. Hence, your knowledge must be accurate.
Likewise, if your chatbot provides an answer that leaves out important details, its response will be confusing and likely unusable by the customer or end user. Your knowledge must contain all the steps or all the information the customer needs; it must be complete.
If the chatbot doesn’t have the information that a system or application was changed last week and has new instructions for use, the bot will serve up information that is out of date, leaving the user frustrated and obliged to contact you. This undercuts the purpose of the chatbot, which is to accelerate answers and keep the routine inquiries out of the support center. (Hint: Updating knowledge needs to be part of your release and deployment activities.)
Your chatbot can’t do everything all by its bot self, so make sure your knowledge is ready to help it.
Roy Atkinson is one of the top influencers in the service and support industry. His blogs, presentations, research reports, white papers, keynotes, and webinars have gained him an international reputation. In his role as senior writer/analyst, he acts as HDI's in-house subject matter expert, bringing his years of experience to the community. He holds a master’s certificate in advanced management strategy from Tulane University’s Freeman School of Business, and he is a certified HDI Support Center Manager. Follow him on Twitter @HDI_Analyst and @RoyAtkinson.