How to Reduce Unconscious Bias in the Workplace

Implicit biases aren’t a moral failing, but rather the result of how our brains process information at scale. Workplaces recognize it exists, but the common response to it, unconscious bias training, produces very little change. 

Yet biased decision-making results in less diverse teams (which we know perform worse) and seriously alters the everyday lives of people, influencing their opportunities for hiring, promotions, and pay raises. 

Below, we look more closely at research on bias and identify the structural interventions across the employee lifecycle that actually help prevent and interrupt it. 

Bias hasn’t gone away. 

Whether it’s recency bias that leads a manager to favor more timely accomplishments in performance reviews, or proximity bias that lends in-person employees an edge over remote workers during promotion conversations, bias shows up consistently across every major decision point in the employee lifecycle. 

We hope that most of the time it’s unconscious bias at play, but explicit bias — acting on stereotypes or preconceived notions about certain identity groups — may also be present in the workplace. 

Hiring and recruiting, the first stage of the employee lifecycle, is where bias first enters. At the most discriminatory firms, Black applicants are 24% less likely to receive a callback than comparable white applicants.

Beyond the traditional biases we think of in hiring, a 2023 meta-analysis found that hiring discrimination against less physically attractive candidates is roughly as severe as hiring discrimination based on race or ethnicity — a finding most organizations haven’t built into their bias mitigation strategy.

The promotion process, the main lever for advancement within an organization, is also rife with bias. A 2025 report from McKinsey and LeanIn.Org showed that for every 100 men promoted to management, only 93 women are. When we look at women of color, the number drops to 74. This trend continues through the C-suite: A 2024 study on executive promotions found that women’s promotion probability was 16% lower than men’s “after controlling for educational and employment background, age, function, rank, and firm characteristics.”

Tour Lattice Analytics

Generally speaking, unconscious bias training doesn’t work.

Unconscious bias training isn’t useless, but it’s not a cure-all, either, and research presents several consistent findings. For one, effects are typically on attitudes, or attitudinal, not behavioral. Secondly, studies often rely heavily on self-reported or short-term outcomes rather than hiring, promotion, or pay data.

Another challenge is the sense of fatalism that training may unintentionally perpetuate. Simply put, when training implies bias is inevitable, some workers may take it as permission to stop trying. 

However, a 2025 meta-analysis on the impact of diversity training on workplace behavior found a small but statistically significant overall behavioral effect.

While that is promising, it’s clear that unconscious bias training is not sufficient as a standalone strategy. “A 16-minute training without action and follow-up and process restructure is not going to change the way that we act or react to certain situations or people or strategies or whatever that might be,” said Daniela Herrera, a culture, talent and DEI consultant. “So does it have a place? Yes, 100% — especially to make sure that we are all in the same place about the conversation, but it can never be the solution.”

{{rich-highlight-1}}

Examples of Unconscious Bias

We’ve established that unconscious bias training must be coupled with process change and intervention for actual impact. At the same time, education is important. Employees at every level of an organization should be familiar with different types of unconscious bias and where biases are most likely to show up. Here’s a non-exhaustive list.

Affinity Bias

Affinity bias, also known as similarity bias, is a tendency to gravitate toward people with qualities or attributes similar to ours. Questions meant to gauge “culture fit” are often thinly veiled affinity bias in action.

In other scenarios, despite their best intentions, recruiters and hiring managers may be influenced by these biases, which can preclude them from considering a more diverse set of candidates. 

It’s easy to see how it happens, said Kim Peirano, organizational psychologist and principal consultant at Integrity Catalyst, a workplace culture consulting firm. “Managers go from ‘I like this person,’ to ‘I just feel like they’re the right fit,’ and then kind of ignore data that comes through.”

Confirmation Bias

Confirmation bias happens when people look for information that confirms a pre-established belief and overlook information that goes against it. This type of bias affects how people collect and perceive new information, but also how they recall and interpret past experiences.

For example, if an employee has a preconceived belief that older workers aren’t skilled with technology, they may notice and remember the times their tenured colleagues had questions about new software rather than the times they used it successfully. 

Conformity Bias

Conformity bias occurs when people behave like those around them instead of acting based on their own judgment or critical thinking. In a group setting such as a work meeting, people may feel inclined to agree with others’ ideas even if they haven’t spent time considering the outcomes.

This is also called groupthink, and it can be detrimental to innovation when the majority of people overlook the same opportunities or choose to ignore a glaring issue for the sake of group harmony.

Gender Bias

Gender bias occurs when a person is treated differently based on their gender identity or gender expression. At the office, an assertive woman might be perceived as “aggressive” while a man with the same attributes might be described as “confident.” A 2024 Textio survey found that women receive 22% more personality-based feedback than men, and that feedback is often biased along stereotypical gender lines. 

Companies that inadvertently enable gender bias will risk the engagement levels of women and non-binary employees, discourage them from sharing their valuable ideas, and brand themselves as an unsafe workplace for gender minorities.

The Halo Effect

Coined by psychologist Edward Thorndike in the 1920s, the halo effect is the tendency to let a single positive impression of someone affect how we evaluate them in unrelated areas. So if people think someone is good-looking, they’ll probably also think they are intelligent and charismatic. The opposite, the horn effect, occurs when a negative trait unfairly influences our overall impression of someone.

Managers need to be wary of generalizing an employee’s performance based on one specific characteristic of their personality or appearance. For example, just because someone is charismatic doesn't mean they’ll be a good leader.  

Anchor Bias

Anchor bias is a mental shortcut that leads people to base future beliefs about a person on the first pieces of information they received. For example, someone who shows up late for a first meeting may be labeled irresponsible or unreliable even when future evidence disproves this assessment. 

Ableism

Ableism is bias against individuals with physical or mental disabilities. Ableist beliefs may lead teams to overlook or even reject qualified applicants because of their disability. But the bias bleeds into the workplace after hiring, too.

In a 2024 survey of 10,000 individuals with disabilities, 30% said that people at work have made negative assumptions about their competence. Plus, many workplaces are not designed to enable or accommodate the needs of people with disabilities or those who rely on assistive technologies. 

Body-Shaming

Implicit beliefs about people’s bodies — such as negative attitudes about physical appearance, weight, size, and ability — can affect the hiring and retention of employees with different body types. A 2026 systematic review on weight stigma found that heavier job applicants are selected less often and are offered lower starting salaries than thinner applicants.

Body-shaming rhetoric can spread when colleagues comment on others’ bodies, even if the subjects of those comments aren’t employees of the company.

Ageism 

Ageism is stereotyping against a person based on their age, commonly affecting older people. In an AARP survey of Americans over 40, 41% of respondents reported experiencing ageism at work in the prior three years. This might manifest during hiring (e.g., older candidates being passed over in favor of younger ones) or in performance reviews (e.g., a manager assuming that an older employee is not adaptable).

Ageism can also affect younger workers, who may be dismissed as inexperienced or not taken seriously regardless of their actual capabilities.

Attribution Bias

This bias occurs when people draw inaccurate conclusions about why someone behaved a certain way. A common form is the fundamental attribution error: When someone else makes a mistake, we tend to attribute it to a flaw in their character, but when we make the same mistake, we blame external circumstances — a classic! 

In the workplace, a manager might assume a struggling employee is lazy without considering other factors, like unclear expectations, inadequate resources, or personal hardship, that may be contributing to their underperformance. 

Beauty Bias

Also known as pretty privilege, beauty bias is the tendency to treat conventionally attractive people more favorably. A 2024 study published in Information Systems Research found attractive MBA graduates earn more than their “plain-looking counterparts.” Beauty bias intersects with other biases — including gender and race — and can reinforce inequitable hiring practices and evaluation practices even when decision-makers believe they’re being objective.

Name Bias

Name bias is the tendency to make assumptions about a candidate’s identity, background, or qualifications based on their name, and to act on those assumptions, consciously or not. For example, one study found that employers were 3% to 24% less likely to contact job applicants with perceived-Black names than those with perceived-white names.

Where and How to Intervene

The research on structural interventions to prevent and reduce bias is clearest in four phases across the employee lifecycle: hiring, performance reviews, promotions, and pay. Below we look at each stage and offer actionable intervention strategies for HR leaders. 

Hiring

“Companies go straight to ‘Let’s build a more diverse pipeline,’ before the hiring process itself is equitable — so they’re funneling candidates into a broken interview process,” said Herrera. “The place to start in hiring in particular is making sure that the actual hiring and interviewing process is as equitable and as unbiased as possible.”

Bias enters here and compounds, but introducing structure by way of clear, documented processes can help.

Structured interviews: A 2021 meta-analysis of personnel selection procedures found structured interviews offer the strongest predictions of job performance. Effective structured interviewing requires defining core competencies, developing behavior-based questions to evaluate for them, and ensuring this common set of questions is asked in the same order each time. Finally, reviewers should rate candidates independently with a standardized interview scorecard before discussing as a group. 

“It’s key that everyone rates applicants on the rubric that was created, and that no one speaks to each other until that’s all done. That way you can avoid anchoring on someone else’s opinion,” said Peirano.

Blinding resumes: Blind or anonymized review can meaningfully reduce bias at the screening stage by removing name, school, and other identity cues before evaluators see an application. The seminal 2000 study by Goldin and Rouse demonstrated that conducting symphony orchestra auditions behind a screen increased the likelihood that a woman would be selected. A 2024 working paper by OFCE researchers reinforced these findings, reporting that blind auditions improved impartiality among judges and mitigated their tendency to select candidates whose gender was already in the majority for that instrument. 

Performance Reviews

Performance appraisals are a key focus for bias reduction because they are where many decisions about career trajectory are made. 

“Performance reviews act as multipliers — they affect compensation, promotion, access to projects, and even future job descriptions,” said Herrera. 

Bias thrives in free-form environments. To reduce the chance of bias in performance reviews, train managers on feedback best practices and calibrate ratings. 

Train managers on feedback best practices.

“Managers need to understand what actionable feedback looks like and why feedback is different from an opinion or an observation,” Herrera said. 

The Situation-Behavior-Impact (SBI)™ method is a useful framework here and supports managers in providing feedback that is directly linked to outcomes. SBI feedback describes the situation where the behavior occurred, specifies the observed behavior, and explains its impact on the team, project, or individual goals.

Calibrate performance ratings.

When done well, performance review calibration ensures criteria are being applied universally across the organization. Done poorly, calibration can entrench biased performance ratings and lead to groupthink. 

For well-run calibrations, HR should:

• Select a rating scale and define the criteria for each rating.
• Determine the expected rating distribution.
• Facilitate calibration sessions to mitigate political influence and subtle biases.

Request a demo

Promotions

The glass ceiling is well acknowledged when it comes to women in the workplace, but as it turns out, the “broken rung” — missing out on a first promotion into a management role — could be an even bigger barrier to leadership representation. When women miss out on promotions earlier on in their careers, there are fewer viable female candidates for senior positions as they progress through the leadership pipeline. According to a 2025 report from McKinsey and LeanIn.Org, women make up just 29% of C-suite roles, and less than a quarter of them are women of color.

Document promotion criteria.

To counter this, implement transparent, documented promotion criteria. When managers know what’s required for employees to advance, there is less room for bias-coded judgment to take over. Rather than choosing someone for a promotion because they “feel like the right fit” — a gut instinct that could be informed by bias — managers must evaluate against a defined checklist. 

Use data to monitor.

To keep tabs on pay equity across the organization, turn to the data. Lattice Analytics can surface promotion rate patterns by demographic group, allowing HR teams to identify where disparities are entering the pipeline and to facilitate data-led promotion discussions rather than relying on informal consensus.

Compensation

Pay gaps, like promotion gaps, are insidious because they are the result of accumulated bias at earlier stages. By the time an inequity shows up in comp data, it may reflect years of evaluation differences; unequal access to feedback, development, or mentorship; and pay and promotion disparities that were never remedied. 

For structured interventions, HR teams should look to increase transparency across the compensation process, culminating in salary transparency. 

But Herrera noted that’s not the first step. “I wouldn’t recommend publishing salary bands immediately; first audit the current pay structure, check for inequities, correct disparities, then communicate transparently. If salary data goes public before those corrections, inequities can harm morale.” 

Conduct pay equity audits (PEAs).

Ideally, HR or an outside firm should be conducting annual pay audits to look for pay discrimination or pay equity issues. Cassandra Faurote, founder and CEO of Total Reward Solutions, said these audits should compare “like with like work,” while taking into account differences for years of experience, professional training and certifications, and education.

Download now

Correct with pay adjustments.

If you find discrepancies, correct them with pay adjustments. To prevent wage compression, pay adjustments should also account for what it would cost to hire a new individual into the role today. 

Communicate findings.

Bolster trust with transparency by sharing the details of your PEA and corrections you’ve made. Doing so will garner trust with the workforce and help keep the organization accountable. 

What AI Can and Cannot Do

AI for bias reduction is now being used across the employee lifecycle. The upsides are real — AI systems can remove some identity cues, flag biased feedback, and support the enforcement of standardized rubrics. 

For example, Lattice AI checks for potentially biased phrasing in feedback and performance reviews to give managers and HR teams an opportunity to review language before it affects decisions.

But concerns about AI’s potential to perpetuate bias are not unfounded. A 2026 Stanford working paper analyzed four million job applications across 156 employers that used hiring algorithms from a single vendor. The researchers found evidence of adverse impact in hiring decisions for Black and Asian candidates and warned that “Algorithms not only influence hiring at each employer but also link outcomes across employers due to shared dependence on the same vendor(s).” With many employers using the same hiring algorithms and vendors, any systemic shortcomings can be broadly replicated. 

This research confirms the need to evaluate AI vendors seriously. Herrera recommends treating AI vendor evaluation with the same rigor as any high-stakes HR process. Before adopting a tool, HR teams should understand what problem it’s actually trying to solve, how it was built, and where the training data comes from, including whether it’s real or synthetic and how often it’s audited. 

Equally important is knowing what happens when something goes wrong: If a biased decision surfaces, what’s the escalation path, and what will the vendor actually do about it? “We don’t necessarily have visibility on what happens after we report the problem,” Herrera said. Given how much sensitive employee data these platforms ingest, she argues the feedback loop has to be held to a higher standard.

“Treat AI systems like new interns handling sensitive data. You wouldn’t give an untrained intern full employee data access without training and NDAs; similarly, HR must assess and monitor AI vendors,” Herrera added.

Building a Bias-Reduction Culture

Structural interventions matter, but they don’t sustain themselves. The organizations that make lasting progress on bias reduction treat it as a cultural commitment and have someone to own it.

Start with accountability. Who at your organization is responsible for bias reduction, and how is progress actually measured? Without a named owner and defined metrics, even well-designed processes will tend to drift. That accountability should live somewhere with real influence over people decisions, not just as an unconscious bias training program that employees sit through once a year.

Psychological safety is another prerequisite for reducing bias in an org. Employees need to believe they can raise concerns about unfair treatment without risking their standing or their relationships at work. If that trust doesn’t exist, the feedback loop breaks down entirely: Bias goes unreported, patterns go unaddressed, and the people most affected quietly disengage or leave.

Data can help you close that loop. Running employee engagement surveys through Lattice Engagement can surface where people feel they’re experiencing inequitable treatment, giving HR teams something concrete to act on rather than relying on anecdote or assumption. 

“Bias mitigation isn’t one big fix,” Herrera said. “It’s daily, detailed work. It’s never one and done.” That means reinforcing the work through ongoing manager enablement, regular review of people data, and transparency about outcomes. When employees see that concerns lead to real changes, trust compounds. When they don’t, it erodes, and no unconscious bias training or check-the-box diversity, equity, inclusion, and belonging (DEIB) activity will get it back.

{{rich-highlight-2}}