Workforce analytics at a federal agency A data-driven approach to optimizing how GSA manages and deploys human capital Paul Tsagaroulis, Director Human Capital Analytics Division Office of Human Resources Management U.S. General Services Administration
2 About us The human capital analytics function is under the Management & Performance Improvement Division in the Office of Human Resources Management, and was established April 2014 The division is taking a data-driven approach to optimizing how GSA manages and deploys human capital to execute business strategy
3 Today s call Topic: Workforce analytics Approach: Data management, visualization Deliverable: Interactive, self-service dashboard Next steps: Forecasting, predictive analytics Additional resources: GovLoop articles, blog by Matt Albucher govloop.com/community/blog/gsa-leveraging-data-visualization-manage-workforce/ govloop.com/community/blog/customers-view-data-dashboards-enhance-decision-making/
4 Workforce analytics dimensions Source: Deloitte (revised for GSA) 01 - organizational capabilities 02 - data collection 03 - information management & data quality Organizational structure, governance, alignment & staffing of workforce analytics / human capital analytics function Techniques & technologies to collect data elements not yet captured by existing systems Transform data from enterprise systems, shadow systems & 'spreadsheet-marts' into useable, accurate & consistent data repository 04 - data strategy & analytics platform 05 - reporting & automation 06 - visualization Enable analytically-driven decision making: determine data needed, how to make it available, & tools to derive valuable inferences from it Transform data into standardized presentations to deliver insights; rapidly & repeatedly recreate analytic products reducing required resources Use proven design techniques to create impactful visual displays of information that effectively communicate findings 07 - ad hoc analysis 08 - rapid technology bridge solutions 09 - advanced analytics Create analytic products in response to one-off, quick turnaround requests Build & maintain tools to support or improve operations until enterprise system can provide functionality Statistical methods, algorithms, or data mining techniques for pattern recognition, optimization, predictive modeling/forecasting of large data sets 10 - sentiment analysis 11 - performance metrics 12 - talent program evaluation Develop, launch, analyze surveys across talent lifecycle, and creating results-based action plans Identify & track key performance indicators to continuously measure progress & evaluate improvements Assess effectiveness of talent programs or projects by gathering & evaluating performance data against measured or desired outcomes
5 Project focused on these dimensions 01 - organizational capabilities 02 - data collection 03 - information management & data quality Organizational structure, governance, alignment & staffing of workforce analytics / human capital analytics function Techniques & technologies to collect data elements not yet captured by existing systems Transform data from enterprise systems, shadow systems & 'spreadsheet-marts' into useable, accurate & consistent data repository 04 - data strategy & analytics platform 05 - reporting & automation 06 - visualization Enable analytically-driven decision making: determine data needed, how to make it available, & tools to derive valuable inferences from it Transform data into standardized presentations to deliver insights; rapidly & repeatedly recreate analytic products reducing required resources Use proven design techniques to create impactful visual displays of information that effectively communicate findings 07 - ad hoc analysis 08 - rapid technology bridge solutions 09 - advanced analytics (next steps) Create analytic products in response to one-off, quick turnaround requests Build & maintain tools to support or improve operations until enterprise system can provide functionality Statistical methods, algorithms, or data mining techniques for pattern recognition, optimization, predictive modeling/forecasting of large data sets 10 - sentiment analysis 11 - performance metrics 12 - talent program evaluation Develop, launch, analyze surveys across talent lifecycle, and creating results-based action plans Identify & track key performance indicators to continuously measure progress & evaluate improvements Assess effectiveness of talent programs or projects by gathering & evaluating performance data against measured or desired outcomes
6 Information management Some of the challenges we faced Customers requested data for their part of the agency but it wasn t practical for HR to support (and anticipate) all the possible views Customers wanted blended reports (onboard employees, who joined or left the agency) but the HR system didn t allow for blending snapshot and transactional data, and there are several different HR reporting applications Business lines described the workforce different from HR but we can t change HR systems, so HR manually adjusted records before sharing The data we reported was often questioned but HR didn t have a process to check and correct bad data before sharing with customers; customers created/managed their own records (risk of shadow HR applications)
7 Data strategy, reporting, automation Challenges Agency leadership wanted reports to aid with data-driven decision making but HR reporting requirements are very different than those of the business lines Customers wanted to see summary data, and line-by-line records but HR had to manually aggregate data and remove personally identifiable information from records before sharing Customers received many different reports, in all kinds of formats but HR seldom standardized formats, and wasn t always consistent with reporting Customers wanted access to real-time data but HR was always reacting to data requests, often bouncing from one fire-drill to the next
8 Visualization, rapid development, analysis Challenges Customers have different preferences for report formats (charts, graphs, tables, line-by-line) but HR couldn t always efficiently meet expectations (could lead to extra work for customers) HR leadership was at the table, but agency leadership wanted timely analysis & reporting but HR didn t always respond with timely, just-in-time analysis Customers wanted analysis, not just reporting but HR spent most of the time preparing report, and very little time analyzing the data; we weren t always asking the right questions HR wanted to jump on the big data train but HR wasn t utilizing the all the data available, reporting was limited to silo applications
9 So, what did we do next? Challenges Deployed workforce dashboard Customer expectations varied, different requirements HR was slow to respond, very reactive -- fire drills Disparate reports; preparing reports was inefficient HR wasn t always providing analysis HR data quality issues, shadow HR systems Duplication of effort at times, decentralized support Conducted focus groups: Met with business partner, HR SMEs to collect requirements and gain consensus Developed data dictionary and HR metrics definitions, developed business rules Promoted transparency, improved efficiency: Sharing agency-wide data in one report Lack of transparency, different data approaches and methods Analytics skills were not utilized, not cultivated Not leveraging workforce data regularly for decision making Developed Tableau expertise: Ease of use, short learning curve; worked with contractor with Tableau technical expertise Tried tell a story with the data, organized report in sections; established one version of the truth
Dashboard: Executive summary
11 Dashboard: Internal labor market Flow of people in / up / out the agency: external joiners, internal promotions, and leavers by grade (also by role, job category)
12 Dashboard: New hires Users can copy charts & underlying data from Tableau into presentations in various formats for use offline
13 Dashboard: Diversity summary Aggregation protects personally identifiable information
14 Dashboard: Appendix Additional filters in the appendix for various drill downs (allows for report customization)
15 Next steps for the project Continue to improve the dashboard Load staffing plans, track vacancies Pre-aggregate data to load to Tableau Use level of detail expressions in Tableau 9: look at an employee record over time (e.g., calculate new hire retention rates, speed-to-competency) Publish dashboard on Tableau Server Provide direct access to agency leadership, heads of business lines, front-line managers Focus on predictive analytics Include forecasts in the dashboard: analyze historical records, and display what can we expect in terms of turnover, and the impact to the workforce
16 Path forward for the analytics function Challenges Opportunities More data than ever before! Start with a theory / hypothesis Disparate data sets, various reporting platforms, need for drill-down reporting Leverage data warehouse environment, with interactive, automated reporting Develop & launch new surveys / data calls for data that are not available Develop scalable, long-term solutions that integrate with analytics solutions Amassed lots of unstructured data Create robust program evaluation metrics, descriptive analysis, predictive analytics, machine learning
Thanks! Contact information Paul Tsagaroulis paul.tsagaroulis@gsa.gov 312-385-3055