How to Solve Hiring Problems with Data Analytics From Data to Insights 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d.
Today s Presenters Casey Johnson, PhD OutMatch Senior Research Scientist Keith McCook, PhD OutMatch Director of Research & Science
A: Um, what do you mean again by Data Analytics? Poll: Do you have experience with Data Analytics in hiring? B: I know enough to be dangerous that s why I m here today! C: I m a Data Analytics Master-In- Training D: I m a Data Analytics Jedi 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 3
What is Data Analytics? Data Structured recording of events Analytics Discovery, interpretation, and communication of meaningful patterns in data 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 4
Description of Data Structured recording of events Always occurs within a context How important is rainfall if there is a source of groundwater nearby? Are you a farmer or a builder? Never without error or bias Error Totally random fluctuation in data Leaf covers opening for 15 minutes Bias Stable distortions in data collection Rain gauge is under a tree Must be interpreted to be useful How does this rainfall compare to other storms this year? How does it compare to previous years? 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 5
Common Data Sources to Solve Talent Problems Turnover Sales numbers Engagement Surveys Performance Reviews Customer Satisfaction and NPS Ratings Productivity Measures Personality Assessment Results 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 6
Description of Analytics Analytics: Doesn t fix anything by itself Must be actionable Data is great, but ultimately useless until analyzed Provides information which helps lead to a clearer understandings of problems Not just numbers and statistics Driven by expertise in the area Is only as useful as it s ability to change the way your company approaches problems Too much Data is almost as bad as not enough. Should support efforts to persuade others 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 7
Know what you are looking for. The more data you have, the easier it is to convince yourself of something that isn't accurate Experts can help Anyone who has experience with a problem similar to the one you are investigating May help identify interesting results that you may want to explore further Replication Gathering more information or similar information on the problem to see if you see the same patterns As data sources increase, your chances of finding false relationships also increase. 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 8
How to set your business up to leverage analytics Structure you data collection to help you answer important questions If you are interested in the performance of employees, using the performance of the store they work in is unlikely to be that helpful Develop hypotheses Choose wisely. Know what you are looking for before you start looking Every good hypothesis should be able to start with the words I bet you $100 that. This can also help decide how to collect information 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 9
A: Not having access to the right data Poll: What is your biggest challenge in Data Analysis? B: Having access to too much data (needle is buried in the haystack) C: Not able to marry the data (i.e., different databases that don t cooperate) D: Not knowing how to analyze (statistical expertise) E: Unclear criteria (unsure what to focus on) 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 10
Using Data Analytics to Solve Problems 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 11
Turnover Turnover is like all-cause mortality We would like it to be lower Understanding how to do that can be difficult Analytics in this case would help get a better understanding of what turnover looks like: Does it happen in a specific set of months? Are there times during an employee s relationship with the company that are particularly risky? Is our assessment helping to resolve turnover issues? 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 12
Financial: Improved Turnover 25 20 15 10 5 Percent Turnover 8% Improvement 14% Improvement 0-30 days 31-59 days Not Recommend (n=479) Recommend (n=1895) Annual Impact 8% improvement in turnover in first 30 days and 14% at the 31-59 day range For the past 12 months, using the Assessment and avoiding those that did not pass would have resulted in hiring 94 people that would have stayed longer. Using a conservative metric for cost of a bad hire: 94 bad hires @ $3500 results in a savings of $329,000 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 13
Energy Company: Impact on Turnover % 35 30 25 20 15 10 5 Percent Turnover 17% change 25% change 4% change 1% change 19% change Annual Impact Approximately 3849 hires a year Assessment demonstrated a 9% improvement in turnover reduction Saved hiring 192 bad hires per year 0 0-30 days 31-60 days 61-90 days 91-120 days 121+ days Termed After... Avoid (n=438) Pursue (n=1166) Cost of a bad hire for a Call Center Team Member is estimated at $3500 per hire ROI: 192 * $3,500= During this 6 month time period, the assessment improved initial turnover then stabled out and later improved long term turnover. Saves $673,680 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 14
Summary of Turnover Analyses Across these analyses we looked at what was occurring at different stages of the employees tenure In the first example we drilled into vulnerable early tenure Effective analysis results Focus on areas of interest In the second we looked at how a suite of assessments impacted turnover across the employee life cycle In both of these cases, with better understanding of the problem, we were able to clearly explain how our assessments were impacting the bottom line. Break it down to understand important broader trends 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 15
Performance Analytics You can gain more insight into what the performance metrics looks like Is tenure having an effect? What levels of performance do most people have? Do you have a few Rockstars driving the effect? Does everybody have consistent performance? You can optimize hiring assessments for key metrics 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 16
Retail: Hiring Sales Associates that Out-Perform! Retail Sales Per Hour Annual Impact $180.00 $175.00 $170.00 10% Improvement $176.89 $176.87 Replacing 34 bad hires gain a $16.28 difference per hour per week 25 equals 1300 hours per year $165.00 $160.00 $160.60 $16.28 X 1300 hours a year X 34 Not Recommends $155.00 $150.00 Not Recommended Recommended Strongly Recommended Hiring only Recommends over a 12 month period results in additional revenue of $719,576! 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 17
Restaurant: Longitudinal Improvements On Server Earnings $20.00 $19.50 $19.00 $18.50 $18.00 $17.50 $17.00 $16.50 $16.00 $15.50 Server Earnings Continued to Increase $19.71 $19.03 $18.33 $18.58 $17.85 $17.07 Not Recommend Recommend Strongly Recommend Server earnings (n=6720) Server Earnings 2017 (n=17437) 2015 Average 2017 Average Continuing to Hire more Strongly Recommends Resulted in Servers raising their average earnings Impact Across 2 Years Those scoring highest on the assessment had large per hour earning differences ($1.18 in 2015 and $2.63 in 2017) Effect: 15,000 servers working 20 hours a week, this is an difference of $25 Million dollars in server earnings from 2015 to 2017 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 18
Summary of Performance Analyses Effective analysis results In our performance analyses we were able to look at the differences in performance between those we recommended and those we suggested the company avoid. Consider the specific impact In the first example we looked how our recommendations effected sales at the hourly level In the second example we looked at how assessment results related to sales at the transaction level, and were able to extrapolate that Benchmark against past experience 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 19
Savings of $2.4M Resulting in a 50 X Return on Investment! 6,143 Applicants 17,375 References Completed 3,300 Hires Average Reference Check 3 days Saved 4,344 Recruiter Hours Resulting in 1,788 Additional candidates! Resulting in $82,603 savings in advertising costs Prevented 660 Bad Hires @$3500 per Bad Hire, $2,310,000 savings $95,563 savings 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 20
Summary of Performance Analyses In this case, simply adding a step to the process resulted in a variety of changes to the hiring process Time saved through using automated reference checking Effective analysis results Candidates added to the system via application by recruiters Ensuring that candidates are likely to succeed based on reference feedback Turn recommendations into $ Although each of these is interesting by itself, the results become more compelling when expressed in dollars. Analysis only make a difference if it can persuade others to action 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 21
Effective Analysis Results Focus on areas of interest Break it down to understand important broader trends Consider the specific impact Benchmark against past experience Turn recommendations into $ 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 22
Summary Analysis provides insight from data Must act on the data for it to be useful Leverage experts to validate data conclusions Can be used to answer many questions relevant to the hiring process. 2 0 1 7 O u t M a t c h. A l l r i g h t s r e s e r v e d. 23