Businesses like chatbots. However, they rarely use the helpers in such a way that they actually relieve them of work. Their development is still often purely marketing-driven, or their potential is underestimated and they only get jobs as avatars with an attached FAQ database. Only a few chatbots are allowed to prove themselves to be real doers. A common consequence: dissatisfaction with the result.
The disillusionment with some chatbot projects is caused by a wrong motivation: Chatbots, which are intended as “eye-catchers” to show that a company is digitally ahead, will eventually lose their attractiveness. Only from the image: “Hello, look, we are relying on artificial intelligence”, key figures such as efficiency and service quality do not improve.
In addition, there is a certain half-heartedness when starting the projects: We see again and again that companies assemble prototypes very quickly, without much planning. They end up in the “PoC cemetery” just as quickly. In other words: They don’t even make it onto the website or into an app, or they are taken offline again after a short guest performance or put on the sidelines – they get a pure FAQ function so that the project does not have to be classified as a bad investment.
Make Chatbots More Involved In Working Life
In my and our imagination, chatbots should relieve employees of their work, for example in sales or customer service, and also offer added value to the customers who use them, and not just chat with them. Companies that want to use the maker qualities of chatbots should pay attention to four central points:
1. Integrate Chatbots Into Business Processes
Chatbots that work separately from the other processes and systems are not helpful. Businesses should let software do a useful job that a human would otherwise do. Users not only want to be guided to a business process by chatbots, they also want to go through it directly. It does not matter whether the user is a customer who places an order or an employee who creates a service ticket. FAQ-based bots do not do any work worth mentioning in such scenarios, as they do not relieve anyone of the workload during the execution of the process, but merely point out to their users.
The central step is therefore to identify a core business process, for example when initiating business or in dealing with existing customers, and then map this completely or in part via the bot. Chatbots can, for example, present products and services if interested parties are initially reluctant to attend webinars with sales representatives. This creates a low-threshold experience with the company – and this is upscaled because chatbots can serve not just one customer, but thousands at the same time. As a process-integrated assistant, the bot can simultaneously record data from potential customers, store it centrally (taking data protection into account) and transfer these leads to sales.
2. Communication Channel And a Targeted Dialogue Structure
Companies shouldn’t use the chatbot as a seller for all sorts of topics. A second important point is therefore to set up the website and chat as a “first point of contact” in such a way that customers are quickly brought to their current destination – and that’s it. Follow-up communication should only be programmed if it leads to a clear benefit for the customer. Cross-selling in the dark is usually a shot in the oven.
A good chatbot recognizes what a website visitor wants with just a few inputs and is directly connected to the relevant business process identified beforehand. If a chatbot in a restaurant is to help reserve a table or put together a menu to be picked up, the software should be excellent in this task and guide the guest quickly through the ordering process to paying.
3. Use The Ability To Learn
Do-it-yourself chatbots learn from user input or should be able to do so. Behind every chatbot there is therefore a worthwhile editing process that should not be underestimated. A lot of knowledge can be gained from the conversation logs and the feedback from the users of the chatbots, which should be incorporated into the further development. With a systematic review, for example, the topic classification – as the chatbot’s understanding of what users want – can be enriched with further examples. This makes chatbots “smarter”. You will need fewer questions in the future and the answers will be more precise. The customer experience improves as a result.
4. Operating Model And User Interface Design
Chatbots have to be operated somewhere. Small and medium-sized companies will rarely set up a local operating platform. The investment will only pay off in exceptional cases. The alternatives are operations from the cloud and microservices that can be quickly adapted to your own needs.
Chatbots will only be busy if they are easy to use. A simple design with a manageable interface is often more appealing for users, who usually use a chatbot when they are in a hurry or on their mobile phone. There is therefore no room for higher aesthetics at the expense of functionality. Customized input and output masks that go beyond the buttons typical of chatbots can, on the other hand, be worthwhile because they enable users not to have to formulate complex inputs using a line of free text. Typical functions of this type are upload functions, selection menus and contact forms. Templates for such masks are available from open source libraries or external providers. This increases user acceptance.