To deliver the right content, chatbots need to be able to access the right content at the right time, quickly and accurately. They need to keep step with the micro-moments that make up our lives.1 What drastically improves this ability is what MindTouch calls Smart content: breaking content into small bites – microcontent – and making it semantically rich. With Smart content, all of the information customers seek that can be found in help content, product documentation, and user manuals is structured hierarchically, which is the way the mind works. This creates an easily navigated decision tree that can lead to the best problem-solving information and heightened customer success.
Search engines favor semantically rich microcontent, making it highly valuable for SEO and especially ideal for mobile sites where parsing a full-length PDF user manual is simply out of the question. Microcontent enables bots that can cross platforms (like Alexa) to easily and instantly pull content directly from a company’s MindTouch-powered website for the user, lowering effort, because it is the most accurate semantically rich content. Breaking out dense help content into bite-sized pieces that are structured to be searchable and machine readable is another way to make bots better.
Further, user activity and inquiries can be tracked, providing rich user data to the company and informing machine learning as the bot matures. And chatbot data analytics are the holy grail for understanding your customers.
Humans vs. machine
Voice recognition in itself is not that new. Eliza was the first stab at natural language processing, in the 60s. Parry came in the 70s; Alice, Cleverbot, and Jabberwacky in the 90s; then SmarterChild.
All along the way we have had the Turing Test to assess how good those bots were at fooling people into believing they were human, which often found significant limitations in the pre-determined responses.2 To date, no chatbot has passed it with flying colors. There have even been bots that deny being bots and try to convince you they are human.
To imitate human capability though is fruitless – machines have nothing on us. They may be super fast, but machines are inherently stupid. Humans on the other hand may have a ridiculously slow CPU in our brains, but we have 100 billion neurons and 100 trillion synapses all performing operations simultaneously and multiple points of input that inform our awareness. We do so many computations without even realizing it – the reality is there is no comparison between humans and chatbots.
Technology has gotten better at replicating real-world exchanges, but the delta is still vast. And maybe chatbots aren’t supposed to exactly be human after all.
Straight to the fix
Alexa can now deliver the answer in response to a query about a Whirlpool product, for example, via software that MindTouch has written that goes and fetches content based on the sentence the customer used. Algorithms connect all of the questions posed by the customers that our current customers serve to all of the answers that we host. These answers are searchable by bots just as they are searchable for search engines, and are compatible with mobile sites. All of this allows companies to reach their customers more directly.
While we structure help content as semantically rich microcontent so that it is easily retrievable and able to show up first in a Google search, now a chatbot does not even have to go through Google. Rather than ask Google, it can ask Whirlpool through MindTouch. Alexa is her own search engine, and MindTouch has curated all of the best answers from the manufacturers’ product documentation and help content so that the customer gets the answer they seek. Alexa can even hold the MindTouch answer for you so that it is available on demand.
Computations are done based on the success of the responses, and that data enables machine learning and aids in anticipating customer questions. When those analytics are processed by MindTouch, any instance where an answer is not available is reported to Whirlpool, whose product team then generates the needed content. So now that Whirlpool knows your unanswered question exists, you and everyone else can get the answer.
- Micro-moments are emerging as a consumer behavior phenomenon that is a result of the larger prevalence of mobile technology usage in daily life. The concept is that our lives have been segmented into hundreds of real-time micromoments, where we consume and act on information after only glancing at our devices.
- See “Chatterbot History: From the First Turing Tests to 2016,” Doky, May 1, 2016; Chatterbot, Wikipedia; and Will Knight, “Tougher Turing Test Exposes Chatbots’ Stupidity,” MIT Technology Review, June 14, 2016.