Machine Learning Times
Machine Learning Times
Three Best Practices for Unilever’s Global Analytics Initiatives
    This article from Morgan Vawter, Global Vice...
Getting Machine Learning Projects from Idea to Execution
 Originally published in Harvard Business Review Machine learning might...
Eric Siegel on Bloomberg Businessweek
  Listen to Eric Siegel, former Columbia University Professor,...
Effective Machine Learning Needs Leadership — Not AI Hype
 Originally published in BigThink, Feb 12, 2024.  Excerpted from The...

9 years ago
Predictive Analytics for Hiring


Wells Fargo uses predictive analytics to hire employees better able to meet its performance requirements and fit into its corporate culture.

“Since its initial launch three years ago, the predictive analytics driven talent assessment solution has been used for pre-employment screening of over 2 million candidates.”

Wells Fargo & Co. has found a more effective way of hiring candidates who will more likely perform better and stay longer – by using predictive analytics in its selection process.

“We thought we should have a more customized solution, as Wells has a very unique culture around needs-based selling and customer service,” says Sangeeta Doss, senior vice president, recruiting manager for community banking.

Doss will discuss in detail how the San Francisco company has been utilizing “enterprise talent analytics” to help recruiters “hire better people faster” at the upcoming BAI Retail Delivery 2013  in Denver. She will be joined by Jim DeLapa, president of San Diego-based Kiran Analytics, in a November 5 session entitled, “Using Data to Increase Predictability of Quality of Hire and Efficiency.”

Biometric Screening

In 2010, after Wells bought Wachovia Corp., the company began to centralize recruitment functions and standardize the selection, performance review and employee incentive processes for its community banking division, Doss says. Her team recruits for Wells’ 6,200 retail branches, call centers and online functions, business banking for customers with up to $20 million in annual revenues, and wealth management for customers with up to $1 million in investable assets. Overall, Doss’ team recruits between 50% and 70% of Wells’ 270,000 employees, including most of those who deal with customers.

To aid in standardizing recruitment across the company, Wells selected Kiran Analytics over two other vendors in a request-for proposals (RFP) process because Kiran was able to develop a customized solution to predict the most qualified candidates for teller and personal banker (platform employee) positions based on their background experiences, career motivation, performance, and life/work skills, according to Doss.

DeLapa says Kiran’s CloudCords Enterprise Talent Analytics predictive model does not depend solely on psychometric assessment tests, “as those are easy for a candidate to game. Instead, we focus primarily on biometric data – things that can reasonably be verified, such as ‘How many jobs have you had? How long have you stayed in those jobs? How many promotions have you had? What is the highest level of education that you’ve completed?’ Using such questions, we were able to build predictive models that show strong and reliable indicators of future performance.”

To develop the 65 questions that each candidate for Wells’ teller and personal banker positions would answer online, Kiran interviewed over 40 retail banking subject-matter experts about the factors needed to be successful in those various roles, DeLapa says. The interviews came from the firm’s workforce optimization projects with dozens of banks, as well as from interviews with Wells’ executives, he says.

Kiran administered its initial assessment to 1,000 Wells’ team members already in those positions – tellers, for at least a year and personal bankers for at least two years. The test group of participants included top-, medium- and under-performers representing a full cross-section of demographics. Kiran then applied the analytics to newer team members, to determine whether Wells’ particular culture impacted performance differently than each employee’s background experience. Kiran found similar correlations for culture and background. The resulting predictive model was then based on balancing the need to choose the top performers while minimizing turnover, DeLapa says.

“If we gave all people with an accounting degree interviews, they may be top performers, but turnover would be off the charts,” he says. “It doesn’t do any good to hire people who might be great, but may not stick around.”

Kiran also made sure the assessments did not inadvertently discriminate protected classes under Equal Employment Opportunity Commission guidelines.

Wells phased in Kiran’s solution by region from February to April, 2012. When job seekers apply for customer-facing positions, Wells invites them by email to complete the online 65-question assessment, which is scored in real time. If candidates score high, then Kiran’s solution automatically schedules them for an interview as soon as they complete the assessment. “This gives Wells a real advantage over their competitors at getting these associates, at a huge cost savings to the bank,” DeLapa says.

Success Indicators

From the start of the rollout through the end of the year 2012, Wells collected roughly 1 million job seeker records and found “very statistically significant differences,” up to 99% confidence levels, in performance metrics and retention rates between those team members that the tool would prioritize for hire and other team members Wells would just hire in the market, Doss says.

Wells also measured the retention rate after each of the hires was on board for six months and found that teller retention improved by 15% and personal banker retention improved by 12%. Wells measured the difference in retention between people hired by the tool who were most likely to succeed versus those not as likely to succeed; in some rural markets where there is not as strong a candidate pool, Wells hires “the best of who is out there,” Doss says.

“This has given us the ability to say that, for those team members we are bringing on board who may not have the in-depth experience and life skills that we would want, we will coach them to help them be more successful,” Doss says. “We can determine if we need to give them a different onboarding experience, a stronger coach, and/or buddy them up with other team members for mentoring.”

The tool has also assisted Wells in defining the best “success indicators” so that it could be more proactive in its recruiting efforts, she says. For example, the best tellers tend to have experiences in financial services, retail telecommunications and hospitality, and they also showed strong academic performance in high school and above. That’s helped Wells source candidates and also helped refine the recruiting strategy in some rural markets where there are not as many candidates with these indicators. The company is now establishing financial literacy classes in some rural high schools “so that when students graduate, they will begin to think about Wells for their career,” Doss says.

By: Katie Kuehner-Hebert
Originally published at

Leave a Reply