What is NLP?
Computer science has advanced such that computers are able to understand more than the traditional programming language through machine learning, deep learning and artificial intelligence (AI). Natural Language Processing (NLP) and Natural Language Understanding (NLU) are technologies rooted in AI which give a computer the ability to understand the human language or unstructured text. They use syntactic analysis and semantic analysis to comprehend language. Syntactic referring to the grammatical structure of how we speak and semantic referring to the meaning conveyed in what we say, which alludes to sentence structure and word interpretation. The human language, unlike programming language, is complicated and varied.
So much of what we say is in HOW we say it rather than in the words used. Traditionally, computers have only been able to understand precise language with no room for error – quite the opposite of human language, but NLP has enabled computers to read, decipher and understand human languages. Its development has been a major step to make computers closer to people.
NLP is used so commonly now that we take it for granted. We use it with Amazon’s Alexa and Apple’s Siri, Google Translate, Sat Nav (a.k.a. GPS to Americans), voice-to-text dictation on our phones as well as things as mundane as spell check, spam filters, suggested search engine keywords and autocomplete. NLP technology is everywhere and it makes our interactions with our devices faster, more convenient and easier.
NLP + CX
As human interaction is the cornerstone of any business, introducing NLP to your CX strategy should be a no-brainer. While there are many ways to add NLP to your business, below are a few suggestions as to how you can add NLP to improve your customers’ experiences and gain added insight.
Sentiment Analysis – If you are capturing the Voice of the Customer in your business through surveys, you should absolutely be using sentiment analysis for advanced analysis of the feedback you are receiving. Sentiment analysis is the tool to help you determine if a comment is positive, neutral or negative. It lets the computer do the hard work for you, so the data can be presented in a manner that’s really easy to digest.
Sentiment analysis can also be used to parse through customer comments on social media so you can get a pulse on what is being said about your brand and service across the internet.
Your ability to have customised, accurate data reflected through NLP is enabled by advanced tools as part of an NLP system. This includes lemmatisation which will bring together different forms of a word so they can be analysed as one like plurals or different tenses (i.e., run, runs, ran, running); disambiguation which identifies the different meanings of the same words (identifying bass as a fish, instrument or sound); part-of-speech tagging which can tell if a word is a verb, noun, or adjective; and named entity recognition (NER) through which you can determine words that can be categorized into groups (i.e., people, places, items). The addition of these features enables you to perform a robust and sophisticated analysis of your customer feedback data set with ease so you can focus on results.
Speech-to-Text – Speech recognition technology has been around for quite some time, but its ability to accurately capture human speech has not always been great. As NLP is more highly developed, speech recognition is becoming good enough to understand various accents, the nuance in language and fast talkers. Introducing speech-to-text technology to your CX strategy means that you can learn more from customer calls because everything can be analysed without playing back a recording. You can easily identify trends and perform sentiment analysis of what is being said, giving you a much larger data set to work with than you usually would have. But this, of course, is only helpful if you are using a tool that is really capturing what your customers are saying. If your system is too archaic and doesn’t include things like punctuation, struggles to recognise similar sounding words, or cannot transcribe in real-time, your ability to perform thorough analysis through further NLP tools will be compromised.
Contact Centre – For many businesses, the contact centre is the epicentre for customer data, so it would be a waste not to gain as many insights as possible there to learn the most about your customers and ultimately improve their experiences. With the introduction of NLP to a contact centre, you will not only decrease the burden on agents but also can gain a more comprehensive understanding of how your customers interact with your business.
With NLP you can cut down on call times by letting the computer gather customer information rather than the agent, tagging multiple topics of conversation rather than the older systems which only allows the agent to identify a single topic at a time. This, of course, will allow agents to take on more calls, minimising the time and cost spent on each customer. NLP will also enable automatic transcription of every call, the content of each can then be analysed robustly through sentiment analysis to identify trends in customer choices or decision drivers through the identification of keywords. In addition, you can track customer sentiment (satisfaction, frustration, etc.) which can help improve general contact centre operations by isolating the repeat issues that customers are facing as well as what they find helpful from agents.
Chatbots & Virtual Agents – With chatbots and virtual agents, NLP allows for the machine to interpret not just the language being written but also the context and intent behind it. Customer support chatbots routinely fail in that they only have pre-generated responses to help customers get to the answer they are looking for, so if a customer asks something outside of what is expected an insufficient answer may be given. Alternatively, they fail because they can’t anticipate what the customer will inquire about next. With the addition of NLP to a chatbot, you gain a more sophisticated virtual agent.
The technology in virtual agents has advanced well beyond what we expect from a chatbot. Chatbots can be annoying and difficult for customers to interact with because while coming close, they often miss the mark when answering customer questions. Virtual agents are rooted in conversational AI and are able to truly help customers just as a human customer service agent can. They can offer guidance or advice and steer customers in the right direction to answer more complex queries. It is highly personalised, which is what customers have come to expect.
NLP does not come without its challenges, but the advancement of NLP technology is only going to continue to progress as it becomes more ingrained in our day to day lives. What may now be an added bonus for companies that can afford to introduce it to their CX strategies will soon become a requirement and more accessible for smaller companies as well as more of a necessity if they wish to keep up.