I like chatbots. I think they’re neat and hold immense promise. Unfortunately, most of the bots I’ve interacted with haven’t met the bar to be considered “useful.” In fact, I would say a majority of chatbot experiences that both myself and others have are quantifiably “bad,” impeding people from getting the help they need, yet these failures haven’t stopped organizational appetites for adopting these solutions to help automate interactions.

To accompany our recently released report, “Best Practices For Help Desk Chatbot Success,” and to ensure that the next wave of chatbots are successful, I wanted to break down some of the common underlying causes we’ve seen that have led to these failures. While it would be easy to blame the underlying technology or chatbot platforms (and there have been limitations here, don’t get me wrong), more often the root cause of these failures was actually related to the strategic priorities of those implementing these solutions.

The three common foundational stumbling blocks we see as part of chatbot initiatives are:

  • Poor scoping or treating the chatbot as a science project. Often, chatbots have been brought into organizations as a science experiment or curiosity — something an employee brought in to fool around with or a team used to solve a specific use case. Subsequent scope creep leads to a solution that lacks adequate training or capabilities getting released to end users. Lack of user-facing messaging on what the chatbot can do leads to it being inundated with requests it’s not prepared for. When the chatbot can’t help or provides a poor experience, users abandon the solution, and in turn, the chatbot is shut down. Ensuring that chatbots have the commitment and proper scoping they need to succeed is step one in ensuring longer-term success.
  • Underfunding the chatbot initiative. Similar to treating chatbots as a science project, often chatbot initiatives are brought in to solve massive problems — like automating help-desk requests — without the resources needed to achieve the objective. Skimping out on funding for tooling, not assigning staff to long-term optimization, and committing the bare minimum in development efforts to these tools combine to create a recipe for failure. Unfortunately, there is no “free” here. Even open source solutions will require investment, through development time and talent. Securing ongoing funding for chatbot initiatives will allow you to spend the resources you need in implementing a successful initiative.
  • Prioritizing efficiency over experience. Naturally, when talking about chatbots, what most people want to hear are the business outcomes: the number of tickets the chatbot can deflect, the number of FTEs it can replace (the actual number here is usually zero), the solutions derived from successful lead generation capabilities … the list goes on. And it’s worth noting, organizations have achieved impressive results with chatbot solutions, in some cases deflecting 60% of inbound tickets. When pursuing these outcomes, however, a critical element is overlooked: user success. The ultimate determining factor in long-term success is people actually using your chatbot. In addition to prioritizing business outcomes, ensure that your metrics capture user success, and directly ask your users about their experiences. Users having good experiences with your chatbot will keep coming back, and users defaulting to your chatbot as a mechanism for interacting with your organization will lead to those longer-term business outcomes.