Problem
Workiva has an application called Wdesk. At the moment Wdesk users utilize a search engine to search for help articles when they are have a problem with the app. However, the search bar is not the most effective at listing specific help articles based on the needs of the user. Searching for a topic requires browsing through many different articles, and then browsing through the contents of each of those articles to hopefully arrive at the solution to the user’s problem. Workiva would like to automate the help article search process by tracking the user’s actions while using Wdesk and using this data to provide a help article recommendation.
Solution
We will be using machine learning and deep learning models to predict relevant help articles based on user behavior data. The models will be developed in Python using scikit-learn and keras with tensorflow backend. Our best model will be able to run on AWS and integrated with the Wdesk application.
Purpose
Currently there are no automated tools that suggest relevant help articles to the Wdesk user. Therefore, if a user cannot troubleshoot an issue they will often call a customer support number. With this current setup, if Workiva wants to expand their business, they will need to higher additional customer support staff to handle larger accounts. This is not a sustainable business model. Workiva would like to provide a better customer support experience with their Wdesk app, by using predictive models to help the user troubleshoot instead of humans.
Goals
Software side: Develop a customer support model that requires minimum effort from WDesk users to find the correct help article for their problem. The mechanism behind the customer support is to have a model that learns and adapts to user’s interaction on WDesk to detect possible problems that the user is trying to solve and provide the correct help article.
Business side: A high performing predictive model will mean Workiva does not have to hire as many customer support staff. In other words, the model will help Workiva make more money, by saving money on forgoing the hiring of additional employees.