Beyond customer sentiment: How to put NLP technology to work
Natural language processing tools and apps have finally arrived -- but how are organizations putting NLP to work? Here are some possibilities that might not be obvious.
It was a very long wait, anticipating the day when computers would understand human speech without extensive training and error correction. Now, with Alexa, Siri and Google Home, it seems the wait is over. But the implications of a new era of natural language processing go beyond the obvious.
There's a looming reason for the anticipation: Most people conceive natural language processing (NLP) as a human-computer interface -- a new way of interacting with machines -- and nothing more. Images of talking computers and robots from TV and movies have reinforced this limited vision, resulting in a truncated concept of NLP technology and its impact on business. If NLP is about nothing more than using voices, rather than typing to get computers and bots to do things, then there's nothing more to talk about.
Speech is about more than simple query response, however. Our speech carries more information than the words it is made of, and the role of speech in business processes is more complex than we assume.
Beyond sentiment analysis
Given the prominence of social media and the integration it has already undergone with customer relationship management (CRM) systems, the most straightforward application of NLP technology in business is the extraction of customer sentiment. This emerges in keywords extracted from online conversations and comments produced by product/service users, aggregated for analysis to determine how popular the enterprise brand is or isn't, what customers like or don't like, and what they're enthused about seeing next. It amounts to getting the answer to the question "How do customers feel?" and figuring out how that answer informs product decisions to come.
So straightforward and ubiquitous is sentiment analysis that it has become passé -- just another club in the golf bag of most organizations currently using CRM. NLP makes such a simple and obvious contribution in this domain that it seems equally obvious that it's capable of much more.
Teaching better teaching
One such contribution can be found in a seldom-considered area: training.
Most organizations, from private companies to colleges to service organizations, place a high premium on training and teaching. There is a vast industry of professionals servicing this need, but all too often, people with no training in presentation are thrust into the position of standing in front of a group and teaching something they know -- even if they aren't very good at the teaching part.
NLP technology can help. Analyzing the speech or transcripts of both successful and unsuccessful trainers, then comparing the two to isolate the patterns and practices that lead to good results, can be a meticulous but perfectly doable project. It's fine to study the tips of good speakers and practice what they preach, but it's even better to dig down into the specific keywords, phrases, emphasis and content structures of a particular topic or instructional domain for what works best. Moreover, NLP analysis of the speech of trainers-in-training can quickly identify mistakes or bad habits, making them easier to correct.
And here's an unusual example from U.S. medical schools: All doctors-in-training are required to walk through graded mock presentations to patient stand-ins, who are coached to ask naïve questions. This allows the medical students to become proficient at explaining complex procedures, outlining detailed recovery regimens and delivering bad news with empathy. This process is expensive, time-consuming and too subjective, in many cases. NLP can improve its efficiency considerably.
Additionally, NLP technology can help improve the quality of job interviews. Many employees find themselves in the position of interviewing potential new hires or team members, evaluating potential contractors or service providers, or quizzing customers about issues they're experiencing -- all while having no training and little or no experience doing such things. This often results in a botched process that might have turned out better, had that person been coached beforehand. NLP can be a big help with such coaching.
Meeting workplace efficiency measures
What NLP can do to improve a single individual, it can likewise do for a group. The two biggest time-wasters in the corporate world are email and meetings; the latter is worse, because an inefficient meeting wastes not one person's time but many.
NLP technology can do a lot here. Analysis of a meeting transcript can readily identify who is contributing and who isn't, who is simply repeating what's been said and who is offering new information and insights, and who is staying on topic and who is diverting the discussion. Moreover, it can be used to sort out good meeting leaders from those who need improvement -- and, via the analyses described above, coach those leaders to greater success.
Finally, there's a hidden gem in NLP, rooted in social psychology: When we speak, our word choices and patterns of self-expression reveal a lot about who we are. A person's speech and text can affirm their self-confidence, engagement, openness and other important personality traits; conversely, it can also unveil uncertainty, self-doubt or boredom.
The list of areas in the enterprise where this may be applied is substantial -- interviews, team dynamics, corporate culture development and employee sentiment are just a few. This level of NLP analysis isn't easy to achieve, as it requires considerable technological support (machine learning across domains, contextual sampling and considerable data science skills). It is, however, a holy grail for the organization that seeks to fully understand its collective identity and the potential of its workforce -- hard to get to, but rewarding across the board.