CAI Value
As soon as you start hearing about realistic AI of one kind or another — outside of science fiction — you will likely wonder about its value proposition. What value will it provide in practice? There are quite a few potential sources of business value you, as a producer or user, could get from conversational artificial intelligence (CAI).
Here I list the sources of value that I have gathered from myself and others. I group then into the following areas:
Value to the User
What is the value CAI brings to the end user? I’m talking about good CAI, not necessarily all CAI. Here are a few ideas in alphabetical order.
Acceptance. Especially in the domain of therapy, an artificial agent can attract users who fear social stigma from human therapists.
Availability. Available 24/7. No interruptions. No vacations. No sleep. No need to fit an appointment on a calendar.
Collaboration. Meaning, bots can collaborate with themselves. They can remember more, collectively and integrate that learning better than human agents.
Frictionlessness. Light-weight. Easy for user to install. Nothing to download, no app updates required, easy to learn (in theory).
Immediacy. It reduces customer wait times when initiating and completing conversation. Customers spend less time waiting “in the queue.” They get answers to common questions immediately in a chat window instead of waiting for an email, phone call, or response from another channel. CAI can immediately answer specific questions for customers to make them happier. Consequently, fewer cases get logged for support agents to resolve. This immediacy also extends to redirects beyond the bot, itself. Bots can instantly welcome customers with a branded greeting and direct them to resources they need fast.
Inexpensive. Again in activities like psychotherapy, which can get expensive when visiting a human therapist, an AI-based therapist could be much more cost-effective.
Interactivity. More adaptive in data entry and therefore potentially fewer steps to collect information from a user using an interactive dialog than by using a static data-entry form.
Knowledgeableness. CAI can have direct access to authoritative data and algorithms whereas a human agent must perform slow, manual searches to get access to the same data.
Minimalism. UI Simplicity. Reduced time, effort, or tedium in user input or interaction. One statement could replace typing and selecting in a dozen UI widgets. One statement could traverse any number of steps in a call-routing menu tree. There is low bandwidth requirement for the user to operate.
Multifunctionality. Like a brilliant Renaissance Man (or Woman), a single bot can provide multiple services. They can answer questions or execute instructions — often in the same software systems that I human agent would have to execute them in anyway, with seamless system integration via APIs.
Naturalness. Users don’t need to learn so much about the layout and format of an app. They use normal language (in theory).
Omni-channel. Available on a myriad of text and voice based channels, messaging apps, virtual assistant devices.
Mobility. By this, I mean the experience a customer has with a given company’s chatbot can be a seamless experience when traveling from one geographic location to another. The user can start conversation on the East Coast and pick up that same conversation on the West coast.
Personalization. CAI is more personalized than a website, e.g. FAQ page. In fact, CAI can be more personalized than a customer service representative because it can seamlessly integrate with customer data. Personalized to each user. Recommender systems and other modern personalization techniques can be built-in. CAI can provide more options to engage with customers in the manner they choose, increasing user satisfaction.
Scalability. It is cheaper and more scalable than a call center.
Vigor. No fatigue. They automate mundane tasks without complaint and maintain attention which reduces mistakes.
Value to the Provider
What is the value CAI brings to the owning company? In addition to the above value its end users which indirectly increases customer loyalty, here are a few more reasons why CAI is valuable to the owning company.
Ad-spend-enhancing. Chatbots are currently novel enough that users engage with them more as an advertising chanel than other established chanels.
Calming. A chatbot will not escalate an emotional or heated discussion. In fact, I believe that users, at least users like me, would find it hard to get truly angry at a chatbot the way I sadly have gotten angry at a human customer support agent. When I get angry, it’s because I think I have found someone who can be held accountable for my troubles. I don’t think most people would look at an inanimate chatbot and say, “You are the ultimate source of all my troubles today.” If I sit down and think long enough, I know the young filipina on the other end of the phone conversation is also not ultimately responsible for the problems I am currently facing with the business she represents, but I still think she is alive enough to pass along my frustration to whoever is. Sometimes call center employees don’t know how to calm a customer because they get too emotionally involved in the situation. Again, a chatbot can be deliberately programmed to be good at calming down a customer and it will never disregard its training, even if it is having a bad day of its own. Now, writing a chatbot that does not do its job and thereby annoys or enrages its user — now that’s another story we need to write a happy ending to by doing better at creating chatbots.
Collaborative. Seamless collaboration with a real human agents within a chat channel when needed is possible and becoming expected, which is valuable to maintain human-level quality of service at all times.
Cost-reducing. Reduced expense through automation. Automate common customer service requests. Lighter weight UI is easier to develop and much easier to deploy on multiple messaging platforms. No employee turnover. Reduced cost of retraining new employees. Ease of deployment and upgrade for the business maintainer. CAI has the potential to free up resources for more high-value work”.
Elastic. A chatbot well-architected can handle high or dramatically fluctuating contact center call volume. It’s much harder to predict call volume and hire and schedule human agents to do as well with variable volume, and as cheaply.
Data-gathering. Provides more data about customers for the company to learn from in developing its brand, products, services, etc.
Market-expanding. Attract new target customers because it appeals more to younger generations and because it offers another access point for all target customers.
Performance-improving. CAI can perform better than humans: no boredom, no inconsistencies. They also save time for human agents. Since chatbots can deflect easy cases, agents can devote more time to complex issues that require creativity or teamwork.
Eventually essential. There will be a tipping point transitioning CAI from an early adopter market to an essential tool for all companies to interface with their customers. Those without it will look old-fashioned (according to those already investing in the technology.) “As the underlying AI and NLP technologies mature, chatbots will become more common in many industries, increasing consumers’ expectation of their use as a channel ...”
Success Factors
As a bonus, here is a table I created to summarize the paper “Why Do Chatbots Fail? A Critical Success Factors Analysis” by Janssen et al, 2001.
This table shows a list of critical success factors for chatbots as found in papers and interviews. It also lists the number of papers and experts who mentioned each factor plus an aggregate score (geometric mean) based on those two counts and a rank of factor importance based on that aggregate score.
