Artificial Intelligence in business is still a fantasy for many of us. For some, Artificial Intelligence frees humans from dull tasks; for others, it would alienate us and destroy jobs. But what few of us know is that AI and Machine Learning are already very present in our companies, and that we use them every day!
NLP (Natural Language Processing) is one of the most prolific fields of application of AI. It is the branch of Artificial Intelligence which involves understanding and processing human language.
Its uses in business are very numerous: NLP offers a wealth of opportunities to increase productivity, reliability, and make better decisions.
Today, we are looking at one of the most fertile areas for NLP applications: that of the relationship between brands and consumers.
- 1- NLP and Voice of the Customer
- 2- NLP for social network analysis and e-reputation
- 3- NLP & Virtual Assistants
- 4- NLP for automatic content moderation
- NLP and Customer Relations: endless use cases!
1- NLP and Voice of the Customer
How can we better understand consumer expectations in order to transform them into better adapted products and services, and optimize the Customer Experience?
Analyzing customer reviews and messages helps to understand the reasons for their satisfaction or irritation, and to implement appropriate action plans. The Voice of Customer is the tooled process by which the company collects, analyzes, and prioritizes customer needs, expectations and preferences so that its products and services meet them as closely as possible.
The NLP thus makes it possible to process written feedback from customers on all channels (responses to satisfaction surveys, emails, consumer reviews, social networks, etc.), to classify them into themes and sub-themes thanks to a Semantic Analysis, and to identify the tone of the speech (or Sentiment Analysis).
Feedback: Engie improves Customer Experience thanks to NLP
Engie, leader in the Energy sector, centralizes and analyzes Customer Feedback across all channels using Semantic Analysis. “Being able to analyze customer verbatim is essential to enabling us to understand why a customer is satisfied or dissatisfied in order to act and re-engage the customer.” (Jean-Remy Dudragne, Customer Experience and Innovation Manager at Engie).
2- NLP for social network analysis and e-reputation
Forums, blogs and social networks contain a considerable amount of Consumer Feedback, Internet user comments on brands, as well as questions about the products and services offered.
Semantic Analysis technologies can process and enhance these contributions. Even if social networks raise methodological questions of representativeness and even manipulation, the collection and analysis on these channels contributes to competitive benchmarking. Indeed, Internet users often compare the products of several brands; it therefore becomes possible to understand the motivations of non-clients.
Please note: we must be aware that the performance of Opinion Analysis decreases, sometimes significantly, when applied to a heterogeneous mass of documents collected on the Web and social networks.
Indeed, each content collected does not necessarily express an opinion: it can be a tweet, a blog post, a comment on a forum or any other page unsuitable for an Opinion Analysis. : company presentation, Wikipedia article, patent, telephone directory …
It is therefore important to qualify the content before seeking to extract opinions. Besides, just because someone is speaking does not mean that they have an opinion; a tool which tries at all costs to detect an opinion will generate a lot of errors.
But the Semantic Analysis of social media comments remains a great tool for taking the pulse of consumers, detecting trends, or analyzing a brand’s e-reputation.
3- NLP & Virtual Assistants
Automatic Language Processing technologies allow significant productivity gains in the field of Customer Relationship Management, mainly through the implementation of Virtual Assistants.
In fact, it is estimated that 80% of customer requests concern 20% of the subjects (those which are most frequently discussed). By automating – even partially – its Customer Relationship, a company gains in reactivity towards its customers. The interest is twofold: to generate economies of scale; but also allow human agents to focus on tasks with higher added value and provide better assistance to customers formulating the 20% of atypical requests.
Incoming message reply assistants
Incoming email response assistants automatically categorize and qualify the content of emails received by a company’s customer service department, and provide the advisor with an automated draft response. This response can be fed by a knowledge base but also by other information about the customer, drawn from the information system (bank balance, order history, type of contract, etc.).
And how to talk about Virtual Assistants without mentioning chatbots!
Chatbots are a great way to provide a 24/7 Customer Relationship channel. They can provide the customer with a first level of information, and put the customer in touch with a human advisor if necessary. They can also be a real sales interface: it is possible, via a chatbot, to book a train ticket, order a pizza, or make an appointment with the doctor.
Feedback: Orkyn’ adopts a chatbot
Orkyn’, a home health care provider, has set up a chatbot on its site to advise people with loss of dependence and their caregivers. The chatbot understands the questions asked in Natural Language and is constantly improving with the questions asked.
Read the Orkyn’ case study: A chatbot to advise people with lower mobility and their caregivers
The final category of Virtual Assistants are Voice Assistants. They use speech-to-text technologies, which analyze a digital recording of a human voice to transcribe the content into textual form.
You are no doubt familiar with certain uses of speech-to-text: voice dictation, automated subtitles on YouTube, Voice Assistant on your smartphone or your personal computer, connected speakers.
In Customer Relations centers, speech-to-text makes it possible to finely index and categorize the nature of telephone exchanges between the consumer and the telephone advisor. It can also help to push procedures or product sheets adapted to the context to the latter; this can sometimes improve efficiency, decrease the mean duration of treatment (DMT) and avoid escalation to level two support.
Please note: even if technology has recently seen notable progress, many challenges remain: understanding a speaker in a noisy environment, taking into account accents and separating voices in a multi-speaker conversation remain complex problems. Oral is generally more ambiguous than written and in practice obeys less strict grammatical rules.
4- NLP for automatic content moderation
Less known than the previous ones, automatic content moderation tools can be very useful in the field of Customer Relations.
One example is the automatic moderation of consumer reviews, which analyzes the content of a customer review posted on a website, in order to extract any offensive, inappropriate, defamatory content and any personal data.
It is also possible to automatically moderate or “clean up” comments written on customer files in a CRM tool. The CNIL prohibits any comment mentioning a disability, a sexual orientation, a political or religious opinion, or quite simply offensive mentions. A large brand had been pinned by the CNIL in 2015 following an audit. To avoid this, it is possible to carry out an automated moderation.
NLP and Customer Relations: endless use cases!
These are just a few examples of the applications of Automatic Language Processing for the marketing and Customer Relations professions. We could as well have talked about recommendation algorithms, to boost audience engagement or optimize conversions on an e-commerce site; or marketing intelligence tools, which scan the web for relevant information on your market and your competitors. Whenever a task involves reading and understanding written or spoken text, AI can be of service to humans, especially with new Deep Learning technologies.
Discover our other blog posts about NLP use cases: