Chatbots

Use of chatbots to automate customer communication

Chatbots are dialogue systems with which you can communicate through text or language. They are often combined with avatars and enable requests from (potential) customers to be processed automatically on websites or instant messaging systems. In the long term, customer communication can be fully automated.

Prototypes and demonstrators available

Cross-industry deployment

Suitable for SMEs?

State of play

Until around 2016, chatbots were mainly reserved for the big companies, but they are now also found in small and medium-sized enterprises.

The challenge that chatbots are facing at the moment is user acceptance. Many customers still prefer to be advised by a real person. However, as technology becomes more widespread and reliable, this is changing. At the beginning of 2017, according to a Bitkom survey (Bundesverband Informationswirtschaft, Telekommunikation und Neu Medien e.V.) already one out of four Germans to present chatbots.

Technology and deployment

Description of technology

Possible use scenarios

Chatbots take over services from various parts of the company. On the one hand, in direct contact with customers, i.e. in distribution or e-commerce (smart & personal search function, assistance services such as sales promotion advice, immediate answering services, personalised customer support).

Another field of application is recruitment: when contacting candidates, chatbots facilitate communication and relieve staff. In addition to communication tasks, the use of chatbots is also common for organisational tasks, such as scheduling or other routine tasks.

Gradual introduction

Step 1: Capturing knowledge structured

Similarly to other self-service tools, it is important to collect and categorise knowledge in a structured way. In particular, the following distinction must be made:

— should the possible questions on how to access the knowledge be answered directly by the chatbot?

— is there a need for clarification by means of enquiries?

— is it essential to have direct contact with a human advisor?

Step 2: Knowledge based on rules-based work

Initial training of the chatbot

The basis for each information offer is the knowledge base. It stores the actual information in a structured way. Often this is still done in the form of simple tables containing row-by-line Q & A pairs.

However, artificial intelligence does not necessarily have to do this manually, but can also be automated on the basis of existing files or even structured websites.

Step 3: Refinement mechanisms for detailed requests

Continuous training of the chatbot

Within the knowledge base, artificial intelligence is able to identify further formulation variants based on questions already asked. The Language Assistant always looks at how similar the questions are and calculates a match rate for each user input. If automatic assignment is not possible due to too low match, the chatbot will be clear with a query.

Step four: AI’s self-learning improvement of response options

Dealing with customers adds new word combinations to the knowledge database, thus increasing the precision in answering questions. In the event of completely new questions, the handing over of individual questions to staff can also generate specific new knowledge for the chat bot. It is important throughout the learning process that operators are always given the opportunity to monitor the new content in order to prevent false learning progress.

Opportunities for SMEs

Facilitating customer communication

Making service processes more efficient

Discharge of routine tasks

Costs saved

contact

Do you need support in setting up your business?

Use our technology radar to keep a look at the main technologies relevant to SMEs!

Chatbots

Use of chatbots to automate customer communication

Chatbots are dialog systems with which you can communicate via text input or voice. They are often combined with Avatars and enable websites or instant messaging systems to automatically process inquiries from (potential) customers. In the long term, customer communication can be fully automated.

Prototypes and demonstrators available

Cross-industry deployment

Suitable for SMEs?

The current status

Until around 2016, Chatbots were mostly reserved for large companies, but they can now be found in small and medium-sized enterprises.

The challenge that Chatbots are currently facing is user acceptance. Many customers still prefer to be advised by a real person. But, this is changing as the technology becomes more widespread and reliable. According to to a survey conducted by Bitkom (German Association for Information Technology, Telecommunications and New Media) at the beginning of 2017, one in four Germans already could imagine using Chatbots.

Technology and use

Technology description

Possible application scenarios

Chatbots provide services in various areas of the company. On the one hand, in direct contact with customers, i.e. in sales or e-commerce (smart & personal search function, assistance services such as sales promotion advice, instant service to answer questions, personalised customer care).

Another field of application is recruitment: chatbots facilitate communication with applicants and take the pressure off employees. In addition to communication tasks, the use of Chatbots is therefore common for organizational tasks such as scheduling or other routine tasks.

Step-by-step introduction

STEP1 Capture knowledge in a structured way

Similar to other self-service tools, the structured recording and categorisation of knowledge is of great importance. In particular, the following distinction must be made:

— Should it be possible for the Chatbot to answer questions about knowledge retrieval directly?

— Is it necessary to clarify by asking questions?

— is direct contact with a human consultant indispensable?

Step 2: Incorporated knowledge based on rules

Initial training of the chatbot

The knowledge database forms the basis for every information. This is where the actual information is stored in a structured manner. This is often done in the form of simple tables that contains question-answer pairs line by line.

Thanks to artificial intelligence, however, this does not necessarily have to be done manually, but can be automated on the basis of existing files or even existing structured websites.

Step 3: Mechanisms for refining detailed requests

Continuous training of the chatbot

Within the knowledge database, the artificial intelligence is inferred to recognise further formulation variants based on questions that have already been asked. The voice assistant always looks at how the questions are and calculates a degree of similarity for each user input. IF an automatic assignment is not possible because the match is too low, the Chatbot will ask for clarification.

Step 4: Self-learning improvement of answer options through AI

By dealing with customers, new word combinations are added to the knowledge database, which increases the precision with which questions are answered. In the case of completely new questions, new knowledge can therefore be generated for the chat bot by transferring individual questions to employees. During the entire learning process, it is important that the operators are always given the opportunity to monitor the new content in order to prevent incorrect learning progress.

Opportunities for SMEs

Facilitation customer communication

More efficient design of service processes

Relief from routine tasks

Cost savings

Contact us

Do you need support with the introduction in your company?

Get in touch with us!

Keep an eye on the most important SME-relevant technologies with our technology radar!