Chatbots: From “stupid hello-sayer” to real help

22. October 2019

Chatbots and AI are currently very popular topics. No wonder, after all, current technologies enable us to have a meaningful and helpful conversation with a computer better than ever before. For the most part, this already works without it feeling like a conversation with a machine-controlled, “stupid hello-sayer”. But in our eyes this should not be the only approach. The user is not helped if the conversation feels subjectively “natural” and friendly, but the actual question is not answered.

In our Familie und Beruf (Family and Career) roject, we faced the challenge that the platform provides a large amount of information. We wanted to use a chatbot to save the user from having to search through the content for a long time. But the chatbot should also be friendly and helpful. For us, too, this was still new territory at the beginning.
Here are some valuable experiences that we were able to take with us through our work and would like to share.

Everyone asks differently
One teaches the chatbot – e.g. by means of the QnA-Maker from Microsoft – to ask and answer questions. The biggest challenge is not that the chatbot has a certain answer to a certain question. Rather the following points should be considered:

  • Different questions
  • Spelling mistake
  • Upper and lower case
  • Keywords
  • Wrong wording

Real help in finding content
In our information platform Family and Career we offer many addresses and contacts in addition to explanations on specific topics. Both should be offered to the user by the chatbot as possible answers to questions.
This meant that we had to define different answer forms. A question about e.g. child benefit should be answered on the general information page and the corresponding address. But it should also be taken into account that the question can also refer to different topics within the whole content. Therefore, the sensible “nesting” of answers is a good approach when a lot of information is offered.

Correct answers or not
When answering the questions, it is important not only to pay attention to content and appearance. For example, a too long answer should rather be given as a direct link to the corresponding source.

Dealing with wrong answers should of course also be considered. Here you can integrate technologies that tell the chatbot its wrong answers. This gives you the Changs the answer to improve later. In such a case you can also offer the user to contact a real person.

Chatbots as real added value
Our conclusion about chatbots is that they can be a real added value for the user if they are designed and taught from the user’s point of view. Here the focus should be on task adequacy – in the sense of usability. The chatbot has the task to help the user, for example to find information in complex contents. Besides the usual parameters like: an appealing visual appearance, quick reaction time and a minimum of “human small talk, the focus should be on concept and implementation. Only under these conditions will a chatbot be accepted by the user and used accordingly.

A chatbot for you too?

We are happy to support you in giving your website real added value with a chatbot.