When an on-demand legal services company realized that they weren’t able to forecast project demand or the legal talent supply accurately, they brought in phData to help transform their staffing process from the ground up using advanced data science.
In just five weeks, phData leveraged AWS Sagemaker, Power BI, XGBoost, and Prophet to create two effective proof of concept models that would accurately forecast supply and demand for a 12-week horizon.
Already serving nearly 50 percent of all Fortune 100 companies, this legal services provider needed to quickly adapt its staffing processes to keep up with its high-growth trajectory.
They were losing as much as 25 percent of their potential deals due to staffing shortages and did not have an effective way of forecasting supply or demand.
The problem was they did not have the internal skillset to build an effective model on their own. They turned to phData, with whom they had a long-standing partnership for help.
There were two challenges that this client needed help with, understanding the upcoming demand for services and finding the right number of specialized attorneys to handle the work.
One of the initial questions that needed to be answered was does the company have the appropriate data required to build out these models? After a few initial iterations of the model building, it became evident that there was enough data to create a powerful forecasting solution.
By using AWS’s Sagemaker to create capability limits, and visualizing them with Microsoft’s Power BI, phData was able to show that it would not only be possible but could quickly be accomplished.
To build the demand forecast, phData leveraged sales pipeline history data from Salesforce. A variety of indicators were created, like age of an opportunity and lagged number of opportunities opened, that provided business insight for the model to learn from.
A combination of XGBoost and Facebook’s Prophet was used to create the demand forecast.
Prophet was chosen due to its exceptional ability to determine the overall level of opportunities closed whereas XGBoost was leveraged for its ability to learn from lagging factors to create the overall shape of demand.
To create a realistic forecast of supply for available attorneys, phData worked closely with various team leaders across the business to capture all business context. This allowed the creation of a rules-based methodology, which incorporated factors like roll-off date, attrition, and hiring trends.
In just five weeks, phData gave this customer the insights they needed to more accurately understand its supply and demand needs, transforming the staffing process and how it could avoid shortages in the future.
They now have two high-functioning models that provided demand and supply needs for the next 12 weeks with approximately 80 percent accuracy. Having this powerful, predictive process allows for the organization to be more proactive in its staffing planning, allowing them to realize additional revenue.
Looking into managed workflows or other ML solutions for your organization? Learn how phData can help solve your most challenging problems.
Data Coach is our premium analytics training program with one-on-one coaching from renowned experts.