Case 1
Client: One of the largest job portals in the world
Opportunity: Predict profit propensity of resumes to optimize online acquisition / marketing of resumes and resume search
Solution:
- Define the key variables driving revenue and profit: compensation and propensity to get a job
- Develop a predictive model for action taken by recruiters based on details in resume
- Use segmented linear regression model for predicting views from resume details e.g. work experience, industry, compensation, functional area etc.
- Train and test the model on millions of records
- Develop a score based on propensity to be viewed for easier understanding and scaling
- Implemented the model in search and online advertising
Result
- 125% increase in resume views leading to similar increase in revenue and ~220% increase in profits
- 140% increase in resume modification to increase the score driving a virtuous cycle
Case 2
Client: One of the largest job portals in the world
Opportunity: Target customers for value added services of resume creation and modification
Solution:
- Determine the important variables to predict propensity to buy for domestic and international customers
- Assign weights of the variables for the predictive model for buying propensity using multivariate linear regression models
- Calculate bias for different values of the variables metropolitan areas, roles, compensation, years of experience, industry and roles
- Train and test the model proving its effectiveness
- Finalize the calling list and implement it in the sales channel
Result:
- Doubled the revenue to $30 MM USD in a year for the value added service