Published on 12|11|2019 by Aleksandra Adamczyk & Stanislav Rejthar
How Artificial Intelligence does cope in the daily work of the enterprise, how much does it cost to implement it and how can it improve the NPS result? We talk about details with Stanislav Rejthar, CEO of SentiSquare, a company, which processing large amounts of text data based on artificial intelligence.
Stanislav Rejthar: Let’s have a look at how AI is defined in encyclopedias such as www.techopedia.com. Here you can find that „AI is an area of computer science that emphasizes the creation of intelligent machines that work and react like humans”.
In the past, the AI was mostly used by companies that we perceive as „innovators” or „trend setters”. Anyone can easily name firms such as Amazon, Apple, Google, Nvidia or Twitter as those relying heavily on AI. Nowadays, more and more companies dive into this area. Both well-established conservative multinational corporations as well as tiny startups have recognized the potential and undisputed benefits of AI.
AI solutions can be found primarily in the areas with sufficient (which means very large) amount of data or other inputs the AI can learn from and later process automatically. Our daily lives are filled with information to the brim and so are the companies. We can easily imagine AI solutions to be of a huge assistance in sales, marketing, finance, customer service, HR as well as corporate legal departments. Actually, it is getting harder and harder to think about a process in a corporation that could not benefit from an AI deployment.
SR: As I mentioned before, the use of AI is not limited to one or some of the processes in a company. Furthermore I do not believe there are many limitations as to the type of industry where AI can be successfully implemented either. Naturally, industries dealing with large numbers of inputs in any shape and form are more inclined to adopt AI faster than the rest. However, we may not forget that government, universities or non-profit organization can also benefit from AI.
From the perspective of our company that focuses on one of the parts of AI, the machine learning, we see the greatest value added in the area of improving customer experience of companies doing the B2C business and receiving large amounts of unstructured data from their customers (such as banks, energy or FMCG companies, mobile operators).
SR: At SentiSquare we automate and optimize customer care processes using machine learning. Our aim is to save time and effort while making the processes and customer experience smooth. Our AI reaches excellence in text analytics through unsupervised or semi-supervised machine learning, which is a very fancy way of saying that we understand different ways the customers talk in different channels (emails, texts, SMS, call transcripts) across various industries.
This can bring real benefit to the customers. Let me demonstrate that on some specific use cases. Our technology enables our clients to identify unsatisfied customers who give in their messages subtle signs they might consider leaving the client way before they actually do. This gives the client time to build a solid retention strategy and undertake steps to reverse this situation.
We focus on the other party of a conversation – a call center agent - as well. Our solution can identify a not-so-good call and tell the client what it was about and how an individual call agent was performing. Furthermore such call can be automatically referred to a coach who will then know what to focus on during the next training session. This definitely results in improved customer experience.
Other way of improving customer experience is removing the well-known but unpopular IVR machine from the process. Instead, you can simply ask the customer to state what they want. Our AI will understand that and route the call directly to a relevant agent. For us it is quite simple thing to do, but it has a great impact – it both saves time and makes customer’s experience much smoother.
I certainly do not wish to forget to mention that our AI is indispensable when dealing with customers’ feedback. It does not matter whether you have hundreds or hundreds of thousands of customers and whether they give feedback via an email, orally or send it in SMS. Our AI can read and analyze all of this regardless of what language is used. Then we supply our client with a deep insight into customers’ feedback using automated means, so they can adopt the right measures that will lead to improving their NPS score.
SR: NPS stands for Net Promoter Score. At the moment it is one of the most popular indicators used in business to measure the satisfaction (or the lack thereof) of your customers. You simply ask your customers how likely it is they would recommend the company product or services to a friend or colleague. They reply using the 0 – 10 score. Your goal is to have customers answer using the highest scores (9 or 10), as only those are considered to be your Promoters – people likely to actively recommend your product or services to others.
The importance of knowing how satisfied your customers really are has always been high. However now it is of vital importance to any business. According to a recent study by Adobe and Econsultancy, customer experience (CX) is becoming a number one brand differentiator, stronger price or product. This trend is confirmed by hard numbers provided by McKinsey that showed that the cash flow of CX leaders is growing 300 % faster than that of the laggards.
Another way to highlight the importance of CX is to look at the role of retention. Firstly, the rate at which customers leave your business can serve as an indicator of customer experience quality. Secondly, a study by Harvard Business Review stipulates that 5% increase in retention results in an increase in profit of at least 25 % Therefore, it does not come as a surprise that investments into top-notch CX and retention solutions are accelerating globally.
However, achieving better CX and increasing retention more than your competitors is steadily getting harder – elegant dashboards do not suffice anymore. That is why companies worldwide are turning to AI to help. Adobe says that this pressure to keep up resulted in a 50% more companies investing into AI to boost CX in 2018 than the year before.
A crucial component of any AI deployment strategy in customer service is the Natural Language Processing (NLP) technology. As the utilization of structured meta data is becoming standard, a deeper understanding of customer interactions is required. NLP enables to increase service quality ever further and to access insight about root causes of customer behavior. And that is precisely what SentiSquare’s AI provides.
SR: There are several ways SentiSquare helps its clients to increase the NPS score. Firstly, B2C companies typically have thousands of customers. Therefore the customers’ feedback is often so extensive that just to read it through is beyond human’s capability – before you finish reading, the data is already obsolete. However AI can read and sort huge amounts of text in an incredibly short period of time while reaching the near human precision level.
Secondly, after the feedback has been read and sorted, we analyze it. And at text analysis we are even better than humans. We are able to detect and discover topics and issues you would not even have thought about to look for. Also, we work with facts, not impressions. Our machines do not have emotions and are not burdened with prejudice. With us, you can get objective insight in your customer voice and troubles.
All of the steps described above provide our clients with a great ground for evidence-based decision making. The feedback scores tell you what happened, but our text analytics tells you why. The clients can then easily prioritize the areas to take action in.
Let’s talk about a real case. A client of ours, a mobile phone operator, operates a client center, which asks every year 1 250 000 customers for feedback on its services. Approximately 60 % of the customers who replied leave a written comment in their feedback that needs to be read and acted upon (more than 10 000 SMS per month). As it would be nearly impossible to do manually, we have provided them with our AI tool that not only reads the full feedback, but also provides them with churn prediction - we are able to identify critical points in customers' messages and warn the client which are the customers who are in risk of leaving.
SR: Of course I could state that the benefits of AI deployment always greatly exceed its cost, but I avoid such cliché and will try to explain our pricing model in a more down to earth manner.
When we engage a new client, we start with a proof of concept. The outcome of such proof of concept on client’s data is that it shows what is possible to really do in a particular situation. We also propose solutions that can best address their needs. This is done for a fixed price.
The client then has all inputs they need to consider the benefits of our solution as well as calculate the financial gain of such implementation. Therefore if they decide to deploy our solution, they know well in advance how much they can spend on it and still remain profitable. Our products vary from client to client and so do our prices, but roughly we can say that a month’s operation of our solution can cost between 1.000 and 4.000 EUR.
SR: We are strong believers that AI can save time, money as well as manpower. It does not necessarily mean you will lay off people once you have implemented the AI solution. But it does give you an opportunity to transfer them to a more interesting and rewarding tasks. More importantly, you can achieve significant savings on hiring when you reduce your employee turnover. AI can not only eliminate tedious and boring tasks, but it can also support employees during the continuous on the job learning process, which is a must in the current fast paced business environment.
Imagine you are a working at a role that typically has a higher turnover rate – such as a call center agent dealing with tens and tens of questions and complaints every day. You have just joined the company. And then a very upset customer calls you. Our AI solution recognizes what a particular interaction is about and can offer you a script that will not only provide a correct solution to customer’s problem, but also in a wording that will have the most calming effect on them. And you can utilize the know-how of your whole team gained during the thousands and thousands of previous calls with customers even if you are a newcomer. How cool is it to have a support tool like this during your first days at a new job?
We all know that employees tend to stay in the jobs where they feel supported and where they can contribute. And lower employee turnover definitely impacts costs of hiring, as satisfied employees are also more likely to recommend your vacancies to their friends and acquaintances.
SR: At the present time almost all labor markets in countries across the EU are facing similar challenges. In many countries the unemployment rate is on record low, companies advertise many vacancies and there is a strong competition for talent among employers. No wonder employees change jobs much more frequently than 10 years ago. In this context I would say it is highly unlikely that AI will be a cause for people losing their jobs. AI solutions will rather enable the companies to allocate their very scarce resource – its human capital – to the jobs where they can bring more value added. Why waste people on repetitive or strenuous tasks such as reading every email that comes to a company inbox and resending such email to a particular person within the organization? That can be done by AI with the same level of precision and much, much faster. The employee can instead focus on dealing with the content of the email and for example handle a complaint well and in a timely manner while preventing customer’s churn.
SR: I am pleased to report we have. And it is not only our biased perception, the improvements have been praised also by our clients. Let me give you a taste of what our solutions can achieve in a short period of time.
One of our clients, a large multinational retail company, regularly analyzes the NPS score gained based on customer’s feedback provided orally. The customers reply to open questions. Their replies are then written down into an answer sheet. Client’s intention was also to classify the responses into categories to monitor the performance of different departments and branches within the company.
This client receives around 40.000 responses per calendar quarter. It used to take them 3 weeks to manually categorize about only 30 % of the feedback via routine and tedious work requesting 120 man-days per year. Using our AI, we are able to process 100 % this feedback provided in colloquial language, in spite of many typos, and deliver results in only 3 days after the quarterly feedback delivery date. Furthermore we were able to find hidden insight in the data.
It did not come as surprise to us that this particular project won the award for the best Client Experience Measurement at the Clientology conference in 2019, a Czech customer experience conference organized by Clientology Institute.
Please send them to: email@example.com
SentiSquare is a technology company that deals with customer-generated text analysis. As one of the few companies uses artificial intelligence based on principles of distribution semantics, which provides many competitive advantages, such as language independence. The company was founded in 2014 as a spin-off by team of researchers at the Faculty of Applied Sciences of the University of West Bohemia in Pilsen. SentiSquare currently supplies technology to the contact centers of large companies such as T-Mobile, E.ON or Albert.
+420 603 400 124