Deep Language Understanding for Customer Insights

Customer feedback is undoubtedly tricky.

Customers of any business, B2C or B2B, are very vocal. They are generous with their thoughts on social media, in blog posts, via email, and they’re not skittish about complaints either. More so, for every customer who bothers to complain, 26 other customers remain silent, and it takes 12 positive experiences to make up for one unresolved negative experience.

Companies that pay attention to even a fraction are rewarded handsomely – in 2015, Watermark Consulting found that “Customer Experience Leaders” outperform the market, generating returns a full 35% higher than the S&P 500 index.

Still, customer feedback is one of the most under-utilized resources businesses have. Excuses abound. There is too much data. Capturing the structured elements, like the star rating, is good enough. Sentiment analysis and words counts are all that is necessary.

The truth is, it has been prohibitively expensive so far to truly engage with customer feedback. To completely understand what customers are saying it takes a business executive to read through every piece of email, text, tweet or facebook post. A non-scalable business model by definition.

Still, there is hope – things are changing fast with deep language understanding powered by technologies like Calibrated Quantum Mesh.

Meet Jane: Category Manager Extraordinaire

Let’s take the case of Ms. Jane Acajou, Category Manager for Beauty and Personal Care at a relatively small e-Tailer. Products she manages get about 5,000 feedback posts on the platform every month. She needs to keep an eye on the social media discussions for each of the 318 product lines, even when customers are only discussion the products not the e-tailer. These products get a ton of feedback on bigger platforms like Amazon, eBay or Walmart, which are also relevant for Jane.

Customers frequently complain about merchandise to the relevant department(s). This feedback isn’t Jane’s responsibility, but she senses that valuable insights are being lost in those conversations.

Since e-Tailer is a smaller platform, it has to entice vendors in order to get their attention. e-Tailer promises close collaboration in marketing and tries to position itself as a true partner to its vendors, but Jane thinks it can take this further by finding better, more targeted information from the market for their vendors’ products.

Jane’s small team is driven and works very hard, but Jane is frustrated. A human team can work only on anecdotal basis – they can probably browse through a few pieces of text every day. Jane suspects her team members do this only on their shuttle ride to work, as they have no other time. She feels that despite best efforts, she is driving blindfolded.

There must be a better way.

What is Deep Language Understanding?

In their most frustrated moments, managers all over the world wish for a genie who could suck in all pieces of text taken from the all corners of the internet/internal databases/etc. and answer all of their questions quickly and accurately.

To do this, our genie will have to:

Break down all inputs into sentences and clauses, parsing the relationships between entities
Identify the various ideas and bundle them accordingly, irrespective of the nuances of each individual expression
Use its knowledge of language, product and customer-specific information, and ancillary databases to fully synthesize and make complete sense of the data

Human language is complex – the same idea can be expressed in countless ways, and it’s no surprise we haven’t come across such a computer operated genie before.

Fortunately, the lamp has been rubbed. Applications based deep language understanding can identify entities in all possible expressions, and find true insights – accurate, objective, and most importantly, actionable insights.

Using such a genie, Jane can now engage with her data in many different ways:

Jane can finally understand everything that customers are saying, at a tap of a button. The results collect ideas in all permutations to give Jane a relative sense of importance, so she can respond to the most critical feedback first. (see graphic on the side bar)
Jane can tell the genie to prioritize certain kinds of insights, especially those that are counter-intuitive. This gives her razor-sharp focus to make specific improvements to her business.
Jane can let the genie loose on all competitive data, providing her with a comprehensive picture in terms of vendor diligence, e-Tailer’s competitive position, and summarized reviews of products that are not even on e-Tailer’s platform.
Jane can provide invaluable insights to her vendors by filtering all the data for certain questions that they want answers to, in real time – streamlining her team’s operations while giving her vendors a powerful and easy-to-use marketing tool.
Jane can classify all customer interaction as per her team structure so that the right executive automatically gets the right knowledge, again, in real time. This allows for her team to close the loop on customer feedback, both positive and negative, ensuring that customers feel heard, and therefore respected and valued.

Technologies like Calibrated Quantum Mesh do just this – collect and understand all the ideas hidden in a text and process them to mine valuable, previously unreachable insights.

But it gets better. This fundamental capability can be used in all sorts of applications.

Other Applications for Deep Language Understanding

Janes in industries as disparate as pharma and law can also benefit from technology like Calibrated Quantum Mesh.

For example:

Email Classification – every company dealing with consumers, prosumers, suppliers or other stakeholders get hundreds, if not thousands, of pieces of communications. They can use deep language understanding to classify every such communication so that the right person responds to each of them immediately.
Pharmacovigilance – Pharmaceutical companies monitor social media and search trends to help monitor side effects of different drugs – the cost reductions involved with automating this piece of the process could be huge.
HR applications – Filtering job applications via keywords is standard practice now. Imagine how many great candidates have been missed for using synonyms?
Investigative discovery – Businesses in law enforcement using Deep Language Understanding would have a unique ability to identify patterns in text to help prosecute crimes.

Of course, the uses for deep language understanding are endless – much like a real genie, the only limit to application is our imagination.

If you’re interested in learning more, please set up a meeting with our team.