The Key Focus of Attrition Risk Modeling at iQor
In this episode, you’ll meet Andrew Reilly, a data scientist at iQor. He explains the Net Happiness Score’s method and technology, which we covered in episode 13 with William Adams. Andrew’s data science career at iQor has evolved across several functions. It’s currently focused on finding solutions to operational challenges, emphasizing attrition risk modeling to mitigate employee attrition.
The Attrition Risk Modeling Process
When the Covid 19 pandemic went into full swing, it changed our operating model in many ways. The most significant impact was moving more than 14,000 employees to a work-at-home (WAH) model in 10 days. Today, we have more than 20,000 employees in WAH. We quickly realized the need to measure how the employee felt about their job. We realized the need for tools to measure the agent’s overall mood. In other words, the level of their happiness.
As the Lead Data Scientist on this initiative, Andrew and his team developed the Mood-o-Meter. The Mood-o-Meter is based on a weekly survey sent to all agents that asks just one question: How is life going for you at iQor? The agent scores their response from 1 to 5 with smiley faces, one as least happy and five most happy.
Perspective Gained from Mood-o-Meter
Managers review the Mood-o-Meter data each week. Managers know to check in with the agent if an agent responds with one or two for more than one week. In fact, the team discovered that there is as much as a 20% attrition risk associated with those agents who scored one or two for more than one week.
Leveraging the Mood-o-Meter data and other machine learning tools that can gauge the agent’s mood resulted in the creation of the CheQin process. Skip-level meetings were set up between an agent and a manager. These are informal conversations where managers check in with the agent to listen and learn why they scored their mood low on the Mood-o-Meter survey. These open forum conversations center around what’s working and not working. The CheQin meetings are well documented to create a record of the discussion along with the next steps.
Examples of the CheQin Process in Action
One example Andrew shared is about an agent who had challenges with the technology needed to help solve customer issues. The agent was already very proactive with their team leader to improve their performance. In the skip-level meeting, the manager arranged special training on the tool. The agent was very encouraged by the outcome and felt very supported. Additionally, the manager committed to keeping the communication stream open with the agent.
Qool or UnQool
Part of the CheQin process includes measuring how the employee identifies with either being Qool or unQool. After each skip-level meeting, the manager asks the agent if the issue was resolved from unQool to Qool.
Another agent was feeling the pressure of being on the frontline handling customer calls. The manager showed compassion and explained a career path roadmap to boost confidence and show that iQor supports the employee every step of the way. The agent was motivated to continue to do his job well to stay on the journey.
What’s Next for CheQin
Andrew says the data science team at iQor is looking to add more data sources to predict attrition risks by allowing the models to blend with each other, helping to identify when agents are at high risk of leaving.
Because the CheQins are documented thoroughly, the data science team is developing natural language processing (NLP) scripts to review all the text written in the documentation. By adding more digital CX automation, iQor can keep improving the CheQin process and the working environment that enables employees to excel at their job and career goals.
Andrew has fun with his two young sons, ages three and one. He says that his three-year-old son enjoys playing “vroom-vroom” with toy cars while the younger son watches. Andrew also enjoys playing golf once each month, which is a 2021 resolution fulfillment.
Learn more about iQor digital customer experience capabilities.