Retailers were already under pressure prior to the pandemic, struggling to adapt to a growing online world amid a plethora of competitors. The Covid-19 outbreak has accelerated some of these trends, with more people shopping online. Margin pressure has made automation a requirement, not a choice so organizational structures and ways of working must be transformed. Automated customer services give quick response to customer queries, increase speed and efficiency, can handle multiple queries in less time. This will allow customers to stay home and resolve order problems online, no need to visit offline service centers.
What it does
The process triggers on new customer complaint inserted to the database. Based on complaint type, the process performs actions to resolve the complaint and make the customer happy. The process calculates the complaint message sentiment score to identify customers' discontent level. For high discontent level, process compensates customer troubles with a discount coupon for the next purchase. As a result, the customer will receive an email with a solution to the submitted complaint.
How I built it
The process is built from modules (embedded wizards) that are doing specific process workflow. Those modules have their own namespace for variables to avoid variables overwriting, for example: Module name: Get Text Sentiment Score has GTSS_ namespace for variables: GTSS_TextToAnalyze. The process consists of main wizard which orchestrates 6 modules according to the workflow path. The modules have input and output variables (parameters) and a variable for handled exceptions. Exceptions will be passed to the main wizard and sent to the support team with tech details.
Challenges I ran into
Basic selectors were not working for autogenerated and complicated CRM HTML code. Hopefully, Kryon has provides selectors customization for such cases. Combined web automation, scripting (Python), and API automation in one process. Resolved web automation challenges with adding visual detection activities.
Accomplishments that I'm proud of
It is my first experience with retail customer service automation! Designed and implemented process architecture could be easily scaled. Exceptions elevating and handling help the support team with process maintenance. The embedded wizards could be reused for other processes since they have input and output parameters.
What I learned
From Kryon RPA academy training I have learned automation best practices and implemented them in my process. I have created a delivery services API on the Flask framework to demonstrate REST API requests handling in Kryon.
What's next for Customer Complaints Resolver for Retail Business
Adding new customer issue types, as broken order or wrong order received. The process could be used in attended mode as a helper for the customer support manager.