The Hidden Treasures in Language

As the world goes digital, there is a tremendous opportunity to collect data at every step of a customer journey and use it for optimization and usage expansion. That said, there are fewer people-to-people interactions throughout a customer journey, which makes it challenging for organizations to understand their customers on a human level (e.g. motivation, intent, delight, values, decision criteria). Accessibility, time constraints, feelings, and underlying motivations are factors that drive customers’ purchase decisions. How will you understand these decision paths without human interactions? How will you cut through the noise and tap into what makes prospects take action?

Semantic analysis is decades old. Analyzing text from clients to evaluate if they are ‘positive’ or ‘negative’ towards a product, a statement, an experience, and so on, has been done for ages. Tech has moved forward on how to collect the data (e.g. social media, streaming services, chatbots, etc) and streamlining the process to get to some kind of ‘core of the consumer’ (NLP and ML open source projects such as: BERT, AllenNLP, Rasa, LIGHT, Huggingface, Tensorflow et. al.).?

Perhaps it is fair to say that recent years of cost-efficient and scalable data processing tech have fueled production-viable ML, leading to improved outcomes of these proven semantic approaches. We’ve definitely experienced a sprawl in sales and marketing tools, overusing the term ‘AI’ in recent years, and a recent trend of recruiting and people success tools that rely on people’s state of mind.?

But be careful when ‘AI’ is present, prepare to get disappointed. Many solutions claiming AI are based on limited data or based on single models (or not complex enough model clusters) and this of course results in questionable accuracy and limited ability to support use case complexity. Moreover, mostly only the simplest semantic analysis is involved. In semi-conclusion: the quality of sentiment analysis has probably increased but still has the same underlying basic thinking. So, where does this leave us?

Well, I am left with a thirst for more innovation. I remember working on a project in 2013 where a brilliant mind (Chris Poulin) had developed an ML system that analyzed with 80% likelihood if a returned veteran would need acute assistance to address depression and suicide risk. It blew my mind then, and only now do I see a hint of a trend of researchers and early-stage startups (in competitive marketing, recruiting and talent management, and in sales outreach) catching on. The secret is not what you say, but how you say it! The structure of language tells me who you are and the changes in how you say things - e.g. when did ‘great’ shift to ‘fine’ - tells me when to take action.?

My interest spikes these days in cognitive semantics and language nuance shift analysis. For those of you who know me, I have always been somewhat of an NLP-nerd as well as in cognitive psychology. I honestly can’t express how geekily delicious I think this branch of the linguistics and/or NLP area of semantics is to me. The research area is from the 1980s (or even older) but the opportunity I see ahead is tremendous. With cost-efficient and production viable ML in the mix and some recent evolution in various open source NLP communities around simplifying cognitive semantics, I see the playing field of understanding people’s state of mind as well as decision profile opening like a curtain at an oversold theatre show.?

Did you know that the structure of how you use language actually reveals more about what you like in a person-to-person interaction, what your decision process most likely looks like, your core values, how you prefer to learn or be spoken to… It is no longer what you say (positive/negative) that is interesting, but how you say it (e.g. sentence length, grammar, choice of word nuances, order of words or sentences) and how it changes over time.?

Imagine the possibilities! You could more easily cut through the noise, attract customers from your biggest competitors, get the right people hired faster, get customers to take the meeting or respond, and so on. If you actually know what makes them feel engaged or what would make them choose to spend the time, and if they are ready to do so. You could move mountains.?

In a distributed world, and with the Millennial generation on the rise in organizational decision chains, long-term loyalty and deep connections between providers and consumers will be challenging. With all the data people put out in the world, willingly, that we know reveals what their physiological profile is just by the structure; it will be a golden field for entrepreneurs. Who will be the first to harvest that golden field with the use of cognitive methods?