In Part I of our blog series, Why Your Chatbot Will Never Work, we briefly spoke about some artificial drivers of AI adoption, and most importantly, knowing what your particular business problems are that you are trying to solve. As you can imagine, this is foundational in determining if the proposed value proposition from any vendor holds water. If you are being told by your vendor what your business problem is, please do yourself a favor and go back to the proverbial drawing board and come back to the table with that in hand. I cannot stress that enough.
Let’s take a moment to chat about the main “categories” of value that AI, and (in particular) conversational AI, are poised to solve.
Your Conversational AI value categories
This is perhaps one of the most boisterous claims of the industry and will appear on just about every sales deck you will see in your buying cycle. Let’s break down the proposition— a “chatbot” is put in place add to or augment the functionality that a human provides. The grand majority of these functions are answering questions, which, conversationally, they are well equipped to do. I refer to these as informational functions. They can also do other tasks like creating a ticket on behalf of a user, or providing a car insurance claim without holding for, or the sales pressure of, a human, etc. These are actionable functions. The reason this is important is that the latter will almost always require integration with another system. That is important to understand when weighing cost, project length, and the involvement of your internal IT and SME resources to make it happen. More on that later.
Cost avoidance is almost always a common element in every value proposition and will tag along with all other value propositions quite often. After all, if you can’t demonstrate (or at least blindly accept) the return on investment (ROI), it might be a pretty hard budget sell, and vendors are prepared for that with all kinds of farfetched ROI scenarios in the absence of a bonified one. So, what does real cost avoidance look like? One of my first major AI customers was a direct-to-market bedding manufacturer that wanted to provide domestic customer service in the overnight hours when their customers might be experiencing issues or have questions (a novel concept, right?). The goal was to avoid the cost of hiring one or two more shifts of people to provide this coverage. Now you might say, “Well this sounds a lot like workforce augmentation”, and you wouldn’t be wrong. However, I usually differentiate between the two by saying that workforce augmentation is adding functionality or relieving tasks, and in this case of cost avoidance, it is directly replacing the need for a human in that role. Another example of cost avoidance might be leveraging a chatbot integration into a backend system rather than buying another (more expensive) module of that system to allow the user to interface with it.
Servicing the end customer is probably one of the most unique and compelling value propositions, because customer engagement is not completely dependent on ROI or human effort, but rather on providing a service or function that would otherwise not be possible without conversational AI. In these cases, it is all about servicing the end user in a new and compelling manner. This often provides a competitive advantage or efficiency to the buyer. Another real-life example of this was a previous customer of mine who wanted to expand the push notifications of their service-based system to be a two-way interactive experience through the SMS channel. Rather than just getting a one-way notification, the user could ask questions or invoke actionable functions via text. This was a unique approach, and none of their competitors offered it.
The value that matters
Now that we have, at least, defined some of these buckets of value, the question that reverberates is, “Is the value really there?” What questions can you ask your vendor to answer that question effectively? Instead of telling you the questions, I am going to make a bold statement that will narrow all of those questions to one ultimate question.
First the statement— None of these value propositions are the core (and only) value proposition that matters! They might be resultant of the above value categories, but they do not stand on their own. The only value proposition that matters is usability. If pointing and clicking and scrolling in an application is defined as a graphical user interface (GUI), leveraging a chatbot in a messaging platform would be defined as an alternative conversational user interface. Let’s take a peek of some of the primary advantages of a conversational user interface (CUI) over a GUI that may not be inherently obvious:
- It is a familiar interface to the end user. Anyone who has ever texted a friend knows how to use a CUI.
- The learning curve in comparison to any other GUI is non-existent. Users do not need special training or time for familiarization.
- It is omnichannel by nature. This means that it works everywhere on systems that people already use and are present on.
Take a step into the “Wayback Machine” with me for a moment and remember what it was like to go from a command-line interface with computers to using a mouse and pointing at things? It was transformational, as is CUI, if implemented correctly.
Remember, I left you hanging on the “one” question that can be asked to determine value? Here it is—“Can the user benefit from a CUI over a GUI for my use case?”It is possible. If the answer is yes, you have value, and if not, don’t employ it there.
This brings me to my final point of this installment—CUIs are not a silver bullet, some things are just left better to a GUI. I will give you a prime example— advanced filtered search. I am talking about the typical shopping website that allows you to narrow in on your new TV by size, brand, type, resolution, mounting capability, input versatility, price, and other factors. Can you imagine trying to do that on a bot? You could surely have the user answer all those questions, but what if they wanted to modify their search halfway through? It will not be a great experience and is probably best left to the current GUI implementation.
In closing, measure your use case against the value question, and if it makes sense, ask your vendor to implement a straw man of that in their system, or even a time-fixed Proof-of-Concept (POC) to allow you and your team to prove your hypothesis of value.
In part III, we will talk about “The Maturity Mistmatch”. Is your organization too mature for your chatbot or is it the other way around? We will take an honest look at which organizations can benefit most from this tech and explore how mature Conversational AI is today from a technical perspective. We will also touch on the subject of Ethical AI in practice and what that means for your organization.
See you there!