Together with the Košík.cz team, we launched the first version of our contextual chatbot in full operation in just a single month, handling over 8,000 inquiries per month. Customers of Košík.cz can communicate with Markéta in full sentences and thanks to this, the chatbot does not reduce the high standard of customer service in any way.
Košík.cz, a leading player in the online sale of food and complementary products, approached us at the beginning of the COVID-19 pandemic to find new ways to relieve call centre workers in the face of a growing number of customer inquiries.
During the first 30 days, the chatbot successfully already solved approximately 60% of inquiries and thereby made a fundamental contribution not only to managing the Christmas peak. The chatbot transferred the remaining 40% of questions to the operator on the omnichannel Daktela application, where the operator can see the chat history and directly follow up on the customer's communication with Markéta.
After the turbulent fluctuations caused by the start of the pandemic and the Christmas holidays, the chatbot maintains excellent numbers. Currently, it handles 75 % of inquiries, which equates to 8,000 chats per month with an average conversation length of 1.5 minutes.
This just confirms our experience that if the chatbot works well and is properly trained, customers have no problem interacting with it.
call centre manager, Košík.cz
Košík.cz prides itself on the high quality of customer service and its operators are available to customers seven days a week. With the onset of the pandemic, there was an extreme increase in inquiries from day to day that the customer department had to deal with. The solution could have been to recruit new employees, but due to the lack of candidates and the difficulty of training, they decided on a new and innovative solution that could quickly relieve the operators while maintaining a high standard of customer service.
In order for the chatbot to work as it should, our team first had to analyse the existing communication. It took us 4 days.
We spent 10 days identifying conversational scenarios suitable for automation.
Together with the Košík.cz team, we then created a set of questions in the dialogue in 14 days so that the chatbot did not self-loop and was able to lead customers to a successful end of the conversation.
User testing helped us uncover the small remaining details and inconsistencies in 2 days.
The chatbot Markéta, as Košík.cz named it for its customers, was born in 5 days, which flexibly responds to customer questions, is aware of the context of the conversation and due to this, is better equipped to find the right answers.
We continue to work on the Markéta chatbot and plan to develop it in the area of conversational marketing as well. But everything is not only our merit, Markéta is also working on herself - thanks to artificial intelligence, she is constantly learning and improving without our intervention.
In time, our voicebot will also be joining Markéta, which will serve as a replacement for interactive voice response (IVR), to make the work of operators on the telephone line easier. The voicebot not only greets the customer kindly, but can also predict the problem that is probably causing the customer to call, based on their last actions (for example, sending an order).
Get in touch, and we’ll give you specific examples and suggestions on how coworkers can help you.