Data Analytics is the key to any enterprise's success; infact, doing business without the correct data these days isone of the biggest mistakes a company can make. In fact,for customer support, data analytics is even more vital aswe should understand the most common issues that our customers are facing. It is crucial to understand how our customers feel about our products and services and understand our customers' overall sentiment.

Not understanding this may cause a company to miss the mark and fail to develop new markets. And, all this information is generally collected by customer support teams through customer surveys. This team is often the most effective communication between a company and its customers.

A bit about Maxence Bouygues first

Maxence is a well established enterprise cognitive systems expert and engineer who has spent decades innovating and implementing cognitive systems that enhance customer support for some of the largest companies in the world.

At Forethought Technologies Inc., a software company founded by Dropbox, Palantir and Autonomy alumni, where Maxence has been playing a critical role in building the company's flagship product, Agatha Solve, recommends answers to customer support tickets. Some of the tech companies he worked with include Planisware, Capgemini, Thumbtack, Carta, Gusto, and many more.

Armed with massive experience and extensive connections, Maxence has been leading the customer support technologies revolution through his professional engagements. According to Maxence, the success behind each customer support team is dictated by few factors, namely data collection, understanding customer sentiment & CSAT, and decrypting the data collection Mechanism.

Importance of Data collection

Data is generally essential for the decision-making process in any organization, and without it, we will not be able to identify our problem areas and work towards solving them. 

Customer support teams (let's say in the e-commerce space) receive thousands of tickets everyday. This data can be analyzed to understand what inquiries are shipping, what are cancellations, what are refunds, and password resets. This data is then sifted through, analyzed, and sent to the various other departments in the organization. At least, this is what happens in a perfect internal setting.

By default, it's essential to understand that this is not as obvious as it seems to go over every query and understand the key issues. Customer support teams receive many emails from angry customers and have agents dedicated to responding to these inquiries.

So, what happens? If the management doesn't know their most significant issues, according to Maxence Bouygues, an AI customer service software expert, the consequences could be dire. Therefore, it is essential to have correct data for any company to make the right decisions.

Throughout his career Maxence has been building numerous predictive analytics tools to help companies make better decisions, build more useful and satisfying products. The tools help to identify the following metrics and increase usage and repeat visits retention, customer satisfaction (CSAT) and customer lifetime value (CLV), which are fundamental for a company's Growth.

Understanding Customer Sentiment & CSAT

CSAT is a fundamental metric in customer support, which measures customer satisfaction, and it is considered the standard for understanding how customers feel about your customer service (CS). CSAT scores are a great indicator of customer loyalty and brand advocacy, both of which affect your sales. Companies try to understand how well their customer interactions go.

Because if customers are left angry or unappeased, the brand image will go down dramatically. So before making decisions on what needs to be fixed in their organization, they need to understand the data first. It's not enough to know that customers are happy or unhappy. We need to understand the root cause of the problem.

Are my customers unhappy because of slow support? Is it because of a too strict refund policy or the lack of competence of my agents? Or is it about some of my products in particular? These are some of the questions we should be able to answer.

So, understanding customers' sentiment, combined with other analytics such as what problem the type they have, is the starting point to strategic decisions. However, regardless of how important data is for a business, the one primary problem all businesses face is data collection.

Decrypting the data collection mechanism

Even till date, the approach that most companies use is to tell their customer support team that they need this data and manually ask them to tag and label every ticket while also solving the issue. This is an inefficient approach and it may not be accurate.

Maxence as an expert in AI-powered customer service softwares feels that it is vital for companies to understand the human error in customer service is quite common. 

"You can't always trust agents to do this. As I said before, agents work for BPOs, often there's a quite high turnover, and their training has to be relatively fast. So training agents for all these processes are quite difficult, costly, and often the output quality of the data isn't there."

Could deep learning be a solution?

Maxence believes that AI can help companies stay connected with their customers through timely and efficient customer service as such tools can provide companies with predictive insights to elevate their work. While a customer support agent cannot quickly scan previous products and inventory to recommend similar items or services that a customer may like, AI can do that instantly.

Bouygues, who has been working with companies like Thumbtack to help them automate their customer support requests, which has boomed by ten folds since the global pandemic, stated that the use of AI has several benefits. It reduces the customer's wait time and provides predictive insights and helps companies come up with the best possible solution.

Similarly, according to Maxence, natural language understanding technologies trained with historical data (that is, on the unstructured data of customers) can automatically learn patterns. When these neural network-based models have known, they can automatically tag the tickets across any dimension that we want. Some models specialize in identifying problem types; some identify sentiment, language Etc. So when we deploy these solutions in a couple of hours, they literally transform a customer support team. Now, every new ticket that comes in is automatically labeled, meaning that you can perform any analytics you need.

"A good example of this is a solution that me and my team implemented for Thumbtack to flag their sexual assault types of tickets automatically. In the past, they had to flag them manually, so they wouldn't know how bad this was on their business and how these types of inappropriate interactions would degrade their brand image. Now the model detects 6000 of these every month with over 97% accuracy."

Thanks to this, the company has been able to instantly make appropriate decisions regarding customer appeasement, banning inappropriate users of their platform, and their ability to acquire new customers.

Final thoughts

There is no doubt that a deep learning-powered approach to customer success gives your business a revolutionary insight into engagement scoring metrics that are crucial. Implementation of solutions proposed by Maxence Bouygues can help your company boost your customer satisfaction and lead to transformation of the customer support into a source of revenue instead of it being a cost center.