To see merit more clearly: Go Blind!

A big part of Human Capital Management is bias removal.

I am not anti-bias. Biases like a “bias for action” could be great. But like anything that is unchecked, biases can become very anti-merit very quickly.

Even the harmful biases like those based on gender, race, religion etc. are not always intentional or malicious. Often, people are victims of their own biases and it harms them too in ways that are unknown to them. It harms a business when it rewards things that are not net-positive for the business and when it promotes or stalls growth for people which is anti-merit. Most of these are unconscious, and I believe it is the obligation of HCM consultants and custodians of these systems to try to reduce bias as much as possible and bake in counter-balances right in the processes. Often simply bringing information is context or delaying an initial snap judgement (crossing the chasm from System 1 to System 2 thinking) is enough to break the effect of bias.

Studies show that people discriminate CV shortlists just on the basis of the name of the person.

Women often get lower ratings for the same feedback as their male counterparts.

There are many ways to counter harmful bias in HR processes. In this blog, we propose one of those approaches.

For Hiring:

Blind Recruitment. This is something ORC is very good at. It can ensure that you guarantee an interview to people with merit, and this context ensures a counter-balancing of snap-judgements.

For Appraisals:

Instead of subjective ratings like “Exceeds Expectations”, better way is to give a blind review based on just the abstracted achievements and feedback from multiple sources. This means, you do an initial rating based on a BLIND fact based review where multiple qualitative and quantitative ratings and context stories are given before anyone knows who it is for. This can only be done for metrics that can be measured from secondary sources and don’t need the manager to themselves give the rating. For instance, take billability, utilization rate, Customer-Satisfaction Scores, Annonymized project information, testimonials, peer feedback, project manager (or matrix manager feedback) and data on performance collected over a long period. All of this has to be annonymized, and the performance should be separated from the person.

Sounds like a costly, time-consuming process, right? For the first time, I think Generative AI can enable us to process vast feedback which is in natural language, and anonymize it, process it, tag it etc. to really remove bias from the entire process at a very low time-and-money cost.

For Compensation:

Remove names from the compensation discussion, and focus on the output of the performance process to make initial changes. Compensation changes don’t have to be linked just to performance as there are a lot of other factors that may go into these. But doing an initial change blind will create anchors that counter-act any biases that just the name of the person might unconsciously induce into the person. If I don’t know who I am giving more or less money to, I am less likely to have any biases translate into numbers!