ChatBots – An Untapped Goldmine of User Intent

By John Rushing, Founder

There have been a lot of discussions lately around how conversational interfaces can and will interact with users whether it’s through the use of appliances like Amazon’s Alexa or Google’s Assistant, ChatBots on the web, or via phone, SMS or other channels. In most cases, the discussions seem to be centered around how comfortable users will be interacting with a machine and the degree to which an AI/ML-enabled bot can be programmed and trained to interact more naturally with humans.

I believe however that what is lost in these discussions is the gold mine of data that a ChatBot can provide regarding a user’s intent. Specifically, when interacting with a user on a corporate website. Click tracking and other web tools have been around for decades and while they can provide some approximation of the user’s intent when they go onto a website, no matter how good the tools are, they will always fall a little short. On our site, for instance, I am able to see how many hits the site had and how many of these were unique visitors. I can track a users journey through the site and can aggregate that data to see what paths users typically take through the site. If they came to the site via a search engine or other referrer, I can tell where they came from and/or how they found the site. But, unless a user on the site clicks one of the various email links or uses a comment form to ask a specific question, I like most companies, can only guess why they came to the site and what specific information they were seeking. For all I know, the person on the other side of the screen may have been planning a multi-million dollar project that my firm could have assisted with but I’ll never know because no amount of click-tracking is going to infer that for me.

This I believe is where ChatBots come into play. While they can be fun and interesting on a site, one of the things that most tools used to develop bots have in common is the ability to capture the entire interaction with each user. That is, what exact questions or statements did they type or speak, how was each statement classified and what the bot’s response was, etc. This is necessary to allow programmers and administrators to improve the ChatBot’s performance over time but the transcripts of those interactions provide a literal gold mine of data related to the user’s intent. Never before have companies had the opportunity to learn and understand why visitors chose to come to their sites except through analysis of the user’s path through the site and the pages they viewed. A site that uses chatbots to interact with visitors gives a company the ability to learn exactly what information a site visitor is seeking and why. Better still, we can obtain this information in the user’s own words versus some fuzzy inference derived from the pages they visited or other actions captured as they navigated the site.

From a user perspective, a well-designed ChatBot can help them to get the information they need quickly and efficiently while also providing additional context (conversation) to promote interaction with the user. I strongly believe that in a very short time, we will see wide-spread adoption of conversational interfaces. Not because they’re cool or fun for users to interact with but because the data they provide to companies that use them effectively will be as indispensable as style sheets and contact forms are today.

Keep an eye out for our ChatBot, “Ashley” in the near future. She is being housebroken as I write this and will assist visitors on our site by helping them to find the information they need quickly and efficiently. At the same time, “her” interactions with users will provide us with a veritable treasure trove of data which can be mined to better understand our visitors and how we can tailor our services to suit their needs.