Last week, the Equal Employment Opportunity Commission sued a janitorial and facilities management company for its failure to keep adequate recruitment records. The EEOC alleges that the company used “criminal history assessments” in its hiring decisions but chose not to maintain those records in violation of Section 709(c) of Title VII. While the case is interesting because it was initiated through a relatively rare Commissioner’s Charge, it also serves as a great reminder that employers are required to keep employment selection records available for EEOC inspection.
This case got us thinking about the recordkeeping when employers use big data and analytics in the recruitment and hiring process. The EEOC’s requirements are designed to help that agency determine if a particular protected class is disadvantaged by an employer’s testing or other selection procedures. Big data and talent analytics are used in selection more and more often these days. If the EEOC comes knocking on an employer’s door wanting to know about hiring decisions aided by algorithms, will the employer be able to churn out the required information or records? How easy will it be to access the information? After the required recordkeeping time period, can an employer destroy the electronic records and information associated with analytics? How can a buyer of HR tech ensure that the tech will be responsive to the demands of an EEOC review?
We don’t have all the answers to these questions, and suspect that nobody does yet, but here are some tips that may help employers avoid problems with the EEOC if they use big data and analytics in their hiring process:
1. Know how the algorithm works. Big data can be a mystery. It’s like putting HR data into a box, shaking it a bit, and producing an amazing result that makes HR’s life easier. But the EEOC is not going to accept “It’s magic!” as a response to its inquiries about hiring criteria. Employers need to know enough about the software and processes they use to explain what their analytics measure. They also need to be able to produce records showing how the analytics reached their results.
2. Keep all versions of the algorithm. One of the advantages of analytics is that the algorithms can learn and improve almost constantly as new data is entered. This improvement can be great for the employer, but improvements can create an issue when the EEOC asks what the selection procedures and criteria were on a specific date or during a particular timeframe. If the earlier, unimproved algorithm isn’t kept, the EEOC may conclude that the employer is not keeping proper records.
3. When the time comes, destroy what you don’t need. The EEOC requires that employers keep selection records for one year. With seemingly endless storage space (particularly with cloud or SaaS-based tech), it may seem easy and cheap to keep records longer than necessary, but old information can create problems. Just like purging old paper files, employers must be able to destroy unnecessary records.
4. Test the vendor. Vendors of HR tech may not be familiar with the EEOC’s requirements. Find out if your vendor knows the law by asking good questions. Vendors not familiar with anti-discrimination law and EEOC requirements need to get educated. They should be able to support their customers when the EEOC comes knocking. If your HR vendor contract doesn’t specify that the vendor understands and complies with anti-discrimination law, insist that those terms be added to the contract. If the vendor isn’t prepared to support its customers in responding to an agency investigation, look for a vendor who will.
Yes, we’ve been concerned about big data and HR tech for a while. And no, we’re not trying to stop the innovation that we know will improve HR procedures and processes. But as the EEOC and other agencies try to keep up with where technology is taking HR, we hope employers remain mindful of the law and the requirements of the agencies that enforce it.