Lessons from Implementing a Human Capital Analytics Function

The Personnel Testing Council of Metropolitan Washington (PTCMW) is a Washington DC membership organization for practitioners of industrial-organizational psychology and organization science.

The January 2017 speaker was the outgoing PTCMW president, Matt Fleisher, who heads Global Talent Analytics at FTI Consulting, a global business advisory firm with approximately 5,000 employees and annual turnover of about 1,000 consultants. FTI fields an annual employee engagement survey with a response rate typically between 75 and 85 percent of the workforce.

Matt shared lessons he learned standing up the Talent Analytics function at FTI, many of which echo what I’ve heard from other practitioners in both private industry and government.

Key takeaways for practicing talent analytics

  1. Start by focusing on the actual organizational challenges, not the availability of data or preferred analysis
  2. Use the research literature to identify and report the KPIs that will drive strategic business decisions
  3. Use descriptive analytics and predictive analytics to get to prescriptive analytics – prescribing actionable recommendations based on the data and analysis
  4. When communicating analysis to stakeholders, use the following three-step process:

Here’s what. So what? Now what…

Highs and lows from standing up a talent analytics group

Year 1
  • Created the function
Year 2
  • Automated routine tasks using R
  • Linked 360, employee engagement, and turnover
  • Became victims of their own success – too much incoming work led to quality assurance (QA) issues
Year 3
  • Formalized the work intake process so customers were no longer calling the analytics team directly. Instead, requests for HR analytics went to the HR contact center which created a ticket and put the request in the queue
  • Delegated reporting from the analytics group to HR business partners
  • Created more time for quality assurance activities
  • Dedicated more time to planning longer-term strategic, predictive analytic work

Other lessons learned

  • Using 360, linked individual employee turnover to disrespectful treatment from senior leaders
  • Using employee engagement survey results to predict turnover up to 6 months from survey administration
  • Data quality control / quality assurance should occur in the HRIS and not the analytic software – this may take longer on the front end, but prevents future QA issues with products

Tips for creating products that are actually used

  • Write short emails – 3-4 sentences, max
  • Write short reports – 1-2 pages, max
  • For data-savvy users – create drill-down dashboards, but caveat that small n sizes don’t generalize
  • Keep it as simple as possible – it’s okay to use advanced techniques, but don’t show them
  • Manage expectations – a predictive analysis is not a 30-minute job
  • State and be clear about your assumptions – note what can happen if assumptions don’t hold
Lessons from Implementing a Human Capital Analytics Function