top of page
Search

Enhancing Diversity in Recruitment: How Technology Can Help

55 minutes ago
6 min read

According to SHRM's State of AI in HR 2026 report, 39% of organizations have already adopted AI somewhere in their HR function, and 62% are using it elsewhere in the business. That's not early-adopter territory anymore, that's the mainstream. For staffing and recruitment agencies, that shift lands differently than it does for a single company's HR team. You're not screening for one role, you're sourcing across Tech, IT, Telecom, Retail, and Healthcare all at once, often for multiple clients with different definitions of "the right fit."


Which makes the next question a lot more interesting: Now that the technology is everywhere, is it actually making placements fairer, or just faster?

Used well, the answer can be “both.” AI can surface qualified candidates that a manual search would have missed entirely: someone whose resume doesn't use the exact keywords a job description expects, but who's a strong fit anyway. Analytics can show exactly where good candidates fall out of your pipeline before they ever reach a client. 


Bias Doesn't Disappear Just Because You Automate Recruitment

Handing screening over to an algorithm feels like it should remove human bias. It doesn't; it just hides it better. AI trained on years of past placement decisions picks up whatever patterns were already there, including which names, schools, or zip codes quietly got filtered out for reasons that had nothing to do with skill, and for an agency working across industries, that bias doesn't stay contained to one client or one role.

A 2024 University of Washington study found that resume-screening tools can rank candidates differently based on perceived race and gender alone. Garbage in, garbage out: flawed training data doesn't get corrected by AI, it gets scaled by it. 

A 2026 Stanford study went further, finding that roughly a quarter of Black applicants and 15% of Asian applicants applied to jobs where the screening algorithm produced discriminatory outcomes against their group. For an agency, that risk compounds fast: if you use the same vendor across multiple clients, one biased tool can quietly follow a candidate from placement attempt to placement attempt.


Bar chart from Stanford HAI's 2026 study showing 26% of Black applicants and 15% of Asian applicants applied to jobs where AI hiring tools produced discriminatory outcomes against their racial group.

                                                       Image Source


A few things actually move the needle here:

  • Audit the training data: Check whether it reflects diverse backgrounds before you trust what it spits out, especially if you're using one tool across very different client industries.

  • Audit regularly, not once a year: New York's Local Law 144 already requires annual bias audits with public reporting. Quarterly checks on a dashboard get you ahead of that, not scrambling to catch up.

  • Score skills, not proxies: Zip codes and graduation years quietly correlate with protected characteristics, and they matter even less in fields like IT and Telecom where skills are easy to verify directly.

  • Demand explainability: If a tool can't say why it ranked someone lower, you can't catch the bias before it becomes a habit.

  • Widen the funnel instead of just filtering it: The real value of AI for an agency is surfacing candidates a manual search would miss, not just processing more resumes faster.

  • Keep a human in the loop: Algorithms don't understand career gaps or lateral moves the way a recruiter does.

  • Vet vendors before you sign, not after a complaint: Ask for real audit results, not just a "bias-free" promise.


Measuring the Funnel, Not Just the Placement

A diversity goal without funnel data is just a hope. Treat your placement pipeline like any other funnel: track pass-through rates by stage for underrepresented candidates across each client and industry you serve, since a sharp drop at one specific stage usually points to a fixable problem rather than something vague and cultural. Benchmark those numbers against industry norms for Tech, Retail, Healthcare, and so on, since a healthy pipeline for one sector may look very different from another.

Then connect placement data to how candidates perform and stay once they're on-site. If diverse placements leave early or don't get extended, the work isn't actually done at the offer. Watch which clients your best candidates from underrepresented groups keep going to versus which ones they don't, because that pattern tells you something too. All of it should live on a dashboard leadership actually looks at, because annual reports get filed and forgotten.


Equity Doesn't End the Day Someone's Placed

Here's a piece of this that rarely comes up: what happens to placed workers once they're on a client's floor. Ordinary time tracking data can surface things a placement report never will, and it matters especially in shift-heavy industries like Retail and Healthcare:

  • Who gets the first pick of overtime, and who's stuck with weekends and holidays

  • Whose PTO or flexible work requests get approved fast, and whose sit for days

  • Who gets extended or converted to permanent roles, and who doesn't

  • Who gets client-facing shifts versus back-of-house work, and who's actually in the room when decisions get made

  • Whether accessibility gets treated as core to a placement, not an afterthought. Agencies often bolt it on late, if at all, and it's as much a part of diversity as anything else on this list

If you want to know whether a client's floor matches what they promised your candidate, look here, not at the placement paperwork.


Trust Isn't Built on a Mission Statement

People trust a process when they can see it, not when they're told to. That cuts two ways.

Clearly explain to candidates when AI is part of the process and what it's evaluating. Give real-time status updates, since nothing kills goodwill like applying through an agency and hearing nothing back. Offer feedback where you legally can, even to people who don't get placed.


Internally, give your recruiters and leadership a live dashboard instead of a quarterly slide deck. Visibility across every client and industry you serve is what turns good intentions into follow-through.


"Culture fit" deserves a second look too. It's a phrase that can quietly become a stand-in for bias if it isn't clearly defined, and it comes up constantly in agency–client conversations. The fix is to push clients to write their culture down, specifically, and put it somewhere public, such as on the website or in the job posting, and not just implied in a debrief call.


Worth knowing: the legal ground here keeps shifting. Title VII, the ADA, and the ADEA already apply to AI-driven hiring, and EEOC guidance has said clearly that employers, not vendors, are on the hook. For agencies, that liability question gets murkier when a client is the one actually deciding who gets hired. The EEOC's own guidance was also pulled from agency websites in 2025 amid a policy shift, so state law is the safer thing to watch right now. Illinois, Colorado, and Maryland already have their own rules on bias audits and consent, and more states are likely to follow.


North of the border, Ontario's rules are already more specific than the U.S. federal picture. The Working for Workers Four Act requires employers with 25 or more employees to disclose in any publicly advertised job posting if AI is used to screen, assess, or select applicants, a requirement that took effect January 1, 2026. The Ontario Human Rights Code already applies to any hiring tool, human-run or algorithmic, and the Ontario Human Rights Commission's own Human Rights AI Impact Assessment, released in late 2024, pushes employers to test AI systems for discriminatory impact before deployment, not after a complaint lands. 


What Separates the Agencies That Actually Change

None of this works on autopilot. Someone still has to read the dashboard, chase down the drop-off point, and treat fairness as ongoing work instead of a box checked at placement. Done right, diversity stops needing a spotlight. It gets built into how an agency sources, screens, and places candidates across every industry it serves, Tech, IT, Telecom, Retail, Healthcare, until it's simply how the agency runs, not something anyone has to announce.


Author Bio

Dean Mathews is the Founder, CEO, and Product Director of OnTheClock, an employee time clock and payroll app he built to help small businesses manage their teams with greater simplicity and confidence. Today, OnTheClock serves more than 16,000 businesses and 160,000 employees through an integrated platform for employee time tracking, scheduling, paid time off, and payroll.


With more than 20 years of experience in the software-as-a-service industry, Dean has dedicated his career to building innovative products that solve real-world problems. He is equally passionate about creating a workplace where people feel supported, empowered, and able to reach their full potential.


A self-taught builder driven by curiosity and purpose, Dean believes the foundation of every successful business is its people, including their talent, passion, and potential. 


When he is not perfecting time tracking, Dean enjoys expanding his faith, spending time with family and friends, and finding ways to make the world just a little better. You can connect with him on LinkedIn.

 
 
 

Comments


bottom of page