We are currently hearing a lot about chatbots. This technology has already established itself in the B2C sector in particular. Most of us have probably had a conversation with one without even realizing it. The technology seems to be mature when it comes to answering questions in large customer care centers that are asked several dozen times a day in a similar form. Chatbots are less common in the B2B sector. Today we talk to Fabian Beringer, CEO and co-founder of e-bot7, about the possibilities and limitations of chatbots in an industrial B2B after-sales environment. – An interview with Fabian Beringer
Mr. Beringer, what do you see as the core benefit of using chatbots in after-sales service?
Today, automation, artificial intelligence (AI) and machine learning are bringing real, lasting change to the contact center structure. Virtual assistants and other tools are already serving customers on a large and small scale and companies are increasingly using AI chatbots. The keyword here is customer experience (CX). Chatbots can not only answer customer queries, but also offer a generally improved customer experience and thus increased customer satisfaction. In addition, the intelligent evaluation of newly acquired customer data creates a more precise understanding of customer needs.
Is every chatbot the same or are there technological differences?
In general, there are 3 levels of chatbots based on technologies of varying complexity. However, since handling a conversation is a very demanding task, not all chatbots can offer the same benefits.
A first level chatbot is rule or keyword based, programmed by handwritten rules to answer questions. The scope of the conversation is limited to a very specific use case, and the chatbot uses Natural Language Processing (NLP) very limited or not at all. It may be that the bot does not understand the questions and cannot answer them if the question patterns do not correspond to the rules. Ultimately, the bot can only answer simple questions and not complex ones. This is sometimes frustrating for customers and hinders a dynamic flow of conversation.
Other chatbot models are so-called call-based and generative modelswhich belong to a sub-area of machine learning and partly use natural language processing (NLP) to process inquiries. Here, customer queries are not only processed by a set of rules and behaviors, but they can draw on entities in the knowledge base and give the impression that you are talking to a human. However, these models are more difficult to train and usually require large amounts of training data.
As chatbot technology is still limited at the moment, we have therefore developed a hybrid solution that we Agent + AI® call. In a hybrid solution, chatbots can be combined with the advantages of call-based and generative models. This allows the bot to act independently and only involve the human agent if the chatbot is unable to answer the question itself.
In which direction will chatbot technology continue to develop?
The technology will continue to develop rapidly. It is currently technologically possible to automate up to 80% of text queries. In our opinion, fully automating 100% of text queries will take a little longer. This would require an enormous amount of data, computing power and new technologies. That's why I believe that the hybrid approach, in which the human employee and AI complement each other, will initially prevail in the next few years. With our hybrid agent + AI® model, we are already automating customer service processes and saving costs. The AI system is constantly learning and supports employees with automated responses and processes. This reduces the agents' workload and compensates for fluctuations in service quality.
What concerns about chatbots do you encounter in your day-to-day business?
When it comes to artificial intelligence, discussions are usually characterized by fear or illusion. Artificial intelligence is often confused with science fiction. To raise awareness, e-bot7 has set up its own user network and offers consulting services. We founders are also involved in the German AI Association, among others, in order to advocate for a human-centered and human-friendly use of AI technologies and to lead society into the digital age.
Chatbots are currently mainly used in B2C service. Why is that?
Traditional customer service is characterized by high costs and manual processes. At the same time, customer demand for faster service via digital channels is constantly increasing. The distribution of the volume of inquiries is not balanced, meaning that support capacity is either not permanently utilized or the processing time for customers at peak times is too long.
This is precisely where companies can start using artificial intelligence (AI) and virtual assistants, as customers today expect a personalized, simple, fast and automated dialogue. Chatbots not only fulfill these customer needs, but also offer a generally improved customer experience and thus increased customer satisfaction. In addition, the intelligent evaluation of newly acquired customer data creates a more precise understanding of customer needs.
The B2C sector is setting a good example here and is already using chatbots in many service areas. Companies in the B2B sector should evaluate and initiate the use of bots as soon as possible.
The variance of customer inquiries is much higher in the B2B sector. The reasons for this are, for example, the lower quantities of a series on the market and the greater complexity of the products. How will chatbots conquer this market?
It is true that customer inquiries in the B2B sector are often more dispersed and more complex, as these questions are asked by specialist partners or dealers, for example, who have a great deal of knowledge about specific products and therefore often want to know precise details about a particular product.
But here too, chatbots can be a valuable contact channel for B2B customers in conjunction with a system connection or in a hybrid solution using a human agent.
At the same time, it is also possible to use a chatbot simultaneously for different customer groups from the B2C and B2B sectors. Among other things, we have realized projects with a trade fair in which exhibitors and stand builders with their questions about logistics and parking space allocation as well as private visitors with inquiries about tickets or travel were able to use the contact channel via chatbot.
Why is it that the technology has not yet established itself to the same extent in the B2B sector?
B2B business works differently to B2C, also in terms of customer communication. The purchase amounts are higher and the decision to buy takes longer. In return, there are often higher switching costs and the relationship between companies and B2B customers is often more intensive and meaningful than relationships with B2C customers.
These characteristics contribute to the fact that the technology has not yet established itself to the same extent in the B2B sector. Due to their intensive and often very close relationship with their customers, B2B companies may assume that a chatbot is not personal enough. However, this is precisely where the hybrid solution described above comes in, as the use of human advisors also ensures personal support around the clock. B2B companies can also use a chatbot to generate leads and acquire new customers as part of a predefined process (pre-sales), for example with our Lead Bot.
How high do you currently estimate the acceptance of chatbot technologies in after-sales in traditional B2B industries such as mechanical engineering?
Chatbot technology is currently still underestimated in after-sales in traditional B2B industries such as mechanical engineering, although it can be put to very good use there too. For example, inquiries about spare parts, information from operating instructions for complex products such as homogenization systems or similar can be answered automatically. And, of course, an automated response to questions about information on additional or similar products and services as well as new products and upgrades in the B2B sector is also useful.
Do you know of any real examples of companies where chatbots are already playing a significant role in after-sales in the B2B sector?
Yes, a large company from the plant engineering sector for the food processing industry, for example. This company uses several chatbots to answer their customers' questions about specific products and generate leads, but also to provide contact information and career information.
We are also working with a manufacturer of electrical installation technology and building systems technology that uses chatbots as a high-quality contact channel for specialist partners and private customers to answer queries about specific products, complaints and other topics.
Let's assume a company wants to get more involved with chatbots in its own after-sales. How would you recommend approaching the topic?
There are a few things you should clarify right at the start. It is best to start by collecting repetitive questions and answers that are frequently asked by customers in the after-sales area and are received by email or telephone, for example. This is where the use of a chatbot can save costs by answering these repetitive queries and relieving employees.
You should also outline the processes that need to be handled in after-sales and can be automated. For example, the processing of complaints or the procurement of spare parts could be considered. Also think carefully about where you want to and can save costs in customer service. Last but not least, you should also determine which additional products or new offers you want to advertise.
In your opinion, what are the most common mistakes people make when they want to get into chatbot technology?
People often forget to appoint a person responsible for managing the preparation, deployment and maintenance of the chatbot. This naturally leads to problems in the long term if no one feels responsible. In addition, with hybrid approaches such as our Agent + AI®, you have to make sure you have enough support agents available to help in all cases where the chatbot gets stuck. This is also sometimes neglected.
Approximately how long does it take until you can celebrate the transition to the first live operation in a pilot project? And how long will it take until a mature solution with significant efficiency levers supports day-to-day business?
Integration is simple and only takes 2 to 4 weeks, and is also relatively flexible with various on-premises and cloud solutions. When you can go live also depends on the use case, industry and company size. Pilot projects usually last 3 to 6 months. The operating phase can then begin.
There are also plug-and-play solutions with predefined processes such as our Contextual Dialogs Editor®that can support day-to-day business from day one. This enables companies to easily automate complex customer service processes without having to write a single line of code. The technology is pre-trained and does not require any technical resources on the customer side.
How long does it typically take for such a project to pay off in the B2B environment?
In the telecommunications sector, we have developed our solution for the O2 Customer Service implemented. It automated more than 60% of all incoming first-level support requests within 2 months, reduced the average processing time by up to 68% and routed more than 8000 tickets per week. In addition, we were able to increase customer satisfaction through shorter waiting times and intelligent routing.
Is there a particularly curious response you have received from one of your chatbots?
We asked our bot the question "Who built you?". The bot dutifully replied: "Probably the best chatbot company in the world: e-bot7"
Thank you very much for the interview!
