The VP of Engineering was frustrated. "We've had diversity hiring goals for three years. We've done the training, posted on diverse job boards, used inclusive language in our job descriptions. Our engineering team is still 85% men and 90% white or Asian. What are we doing wrong?"
The answer was uncomfortable: almost everything. Not because their intentions were bad—they were genuine—but because they were doing the visible things that make companies feel good rather than the structural things that actually change outcomes.
After tracking diversity hiring outcomes at over 90 companies through SmithSpektrum, I've learned that the gap between diversity theater and actual diverse hiring is enormous. Here's what separates the companies that succeed from the ones that just talk about it[^1].
Why Most Diversity Efforts Fail
Let's start with why the standard playbook doesn't work.
Unconscious bias training has minimal impact on hiring outcomes. Meta-analyses consistently show it changes attitudes temporarily but doesn't change behavior[^2]. Companies that rely on training alone see no measurable improvement in hiring diversity.
Diverse job boards help, but only marginally. If your interview process filters out diverse candidates, it doesn't matter where you source them.
Inclusive job descriptions are table stakes, not differentiators. Removing "ninja" and "rockstar" won't fix a broken pipeline.
Diversity goals without accountability become aspirational statements that no one owns. If missing the goal has no consequences, it's not really a goal.
The pattern: companies address the visible parts of hiring—sourcing and messaging—while leaving the invisible parts—evaluation and decision-making—unchanged. The invisible parts are where bias lives.
What Actually Works
Companies that build diverse engineering teams do different things at every stage of the hiring process.
Stage 1: Pipeline Building
The goal isn't just to source diverse candidates—it's to build relationships with diverse talent communities over time.
| Approach | Effectiveness | Why |
|---|---|---|
| Conference sponsorship (diversity-focused) | Medium | Brand awareness, not immediate pipeline |
| University partnerships (HBCUs, HSIs) | High | Consistent early-career pipeline |
| Bootcamp partnerships | High | Non-traditional backgrounds |
| Employee referral incentives (diverse candidates) | Medium-High | Leverages existing diverse employees |
| Apprenticeship programs | Very High | Builds pipeline while reducing risk |
| Open source community engagement | Medium | Technical credibility signal |
The highest-performing approach is apprenticeships. Companies that run engineering apprenticeship programs targeting underrepresented groups see 3-4x improvement in diverse hiring rates. The apprenticeship period reduces perceived risk, allows candidates to demonstrate ability, and creates genuine pathways for non-traditional backgrounds.
Stage 2: Sourcing
Where you look determines who you find.
| Source | Diverse Candidate Yield | Notes |
|---|---|---|
| LinkedIn (standard search) | Low (~15%) | Reflects industry composition |
| LinkedIn (diversity filters) | Medium (~30%) | Intentional sourcing helps |
| /dev/color, Lesbians Who Tech | High (~90%+) | Community-specific platforms |
| Jopwell, Fairygodboss | High (~85%+) | Diversity-focused job platforms |
| Code2040, Year Up alumni | Very High | Pipeline programs |
| Employee referrals (targeted) | Medium (~40%) | Depends on team composition |
The key insight: if you source the same way everyone else does, you'll get the same demographics everyone else gets. Intentional sourcing from specific communities is required, not optional.
Set sourcing targets—not hiring targets—for your recruiting team. "50% of candidates entering the interview pipeline should be from underrepresented groups" is actionable and measurable.
Stage 3: Resume Review
This is where hidden bias has the most impact.
Name, school, and company logos trigger unconscious associations before anyone reads the actual qualifications. Studies show identical resumes with "white-sounding" names receive 50% more callbacks than those with "Black-sounding" names[^3].
What works:
Blind resume review removes names, photos, school names, and company names from initial screening. Tools like Blendoor or internal processes can anonymize applications.
Structured criteria before review means defining what you're looking for before you see any resumes. Write down the specific qualifications that matter, then evaluate each resume against those criteria.
Multiple reviewers reduce individual bias impact. Have at least two people review each resume independently.
Skills-based assessment first means starting with a skills test or work sample rather than resume review. Candidates who pass the skills bar then get resume review—reversing the typical order eliminates the resume as the first filter.
Stage 4: Interviews
Interview processes that feel "objective" often aren't.
| Common Practice | Problem | Fix |
|---|---|---|
| Unstructured conversations | Favors rapport over competence | Use structured questions |
| "Culture fit" evaluation | Favors similarity | Define specific values to assess |
| Brain teasers | No correlation with performance | Use job-relevant problems |
| Whiteboard coding | Advantages CS degree holders | Use realistic environments |
| Single interviewer decisions | Individual bias unchecked | Require multiple perspectives |
The core principle: every part of the interview should be standardized, job-relevant, and evaluated against predetermined criteria.
Diverse interview panels help, but they're not sufficient. Having a woman on the panel doesn't prevent the panel from making biased decisions. Panels need structured rubrics and calibration, not just diverse composition.
Interview training on structured evaluation matters more than unconscious bias training. Teach interviewers to use rubrics, score independently, and distinguish between "different from me" and "won't succeed here."
Stage 5: Decision Making
The hiring decision is where good intentions meet reality.
Require diverse finalist slates. The Rooney Rule—requiring at least one diverse candidate in the final round—has mixed results. A stronger version: require that the finalist slate reflects your pipeline diversity. If 40% of candidates entering interviews were from underrepresented groups, but 0% of finalists are, something broke.
Calibration discussions must address bias explicitly. When debriefing candidates, someone should ask: "Are there any concerns that might reflect bias rather than job-relevant issues?" This makes it safe to name patterns that might otherwise go unspoken.
Track decision patterns by demographics. If diverse candidates pass phone screens at the same rate as non-diverse candidates but drop off at on-site interviews, you have an on-site interview problem. Data reveals where bias lives.
Building Inclusive Interview Experiences
Diverse candidates often face additional friction in interviews.
| Barrier | Impact | Solution |
|---|---|---|
| Unfamiliar with "tech interview culture" | Underperformance on gameable elements | Provide prep materials |
| Imposter syndrome | Self-elimination or hesitancy | Affirm belonging |
| Different communication styles | Misread as "not confident" | Train interviewers on style differences |
| Caregiving responsibilities | Schedule inflexibility | Offer flexible scheduling |
| Accessibility needs | Physical or cognitive barriers | Proactively offer accommodations |
The highest-impact intervention: send all candidates detailed interview prep materials. What to expect, what topics to prepare, what format each interview will follow. This levels the playing field between candidates with interview coaching access and those without.
Retention: The Often-Ignored Half
Hiring diverse engineers means nothing if they leave.
Underrepresented engineers leave companies at higher rates than majority engineers. The reasons are consistent: lack of belonging, limited advancement, microaggressions, and isolation[^4].
| Retention Factor | What It Looks Like |
|---|---|
| Belonging | ERGs, visible leadership representation, inclusive team norms |
| Advancement | Clear promotion criteria, sponsorship programs, manager training |
| Safety | Zero tolerance for discrimination, clear reporting mechanisms |
| Community | Cohorts of diverse employees, mentorship programs |
Companies that focus only on diverse hiring without addressing retention are filling a leaky bucket. The most successful companies track retention rates by demographics and treat attrition gaps as seriously as hiring gaps.
Metrics That Matter
Measure what actually indicates progress:
| Metric | What It Shows | Target |
|---|---|---|
| Pipeline diversity | Sourcing effectiveness | 40%+ URG* candidates |
| Pass-through rates by stage | Where bias occurs | Equal rates across groups |
| Offer acceptance rate by demographic | Candidate experience | Equal rates across groups |
| Time to promotion by demographic | Advancement equity | Equal or better for URG |
| Attrition by demographic | Retention/belonging | Equal or lower for URG |
| Engagement scores by demographic | Experience quality | Equal scores across groups |
*URG = Underrepresented Groups
The most important metric is pass-through rate by stage. If diverse candidates enter your pipeline at 40% but receive offers at 20%, you have a process problem. Track where candidates drop off.
What Leadership Must Do
Diversity initiatives fail without leadership commitment. That commitment must be visible and structural.
Set public goals with accountability. Leadership should state diversity goals publicly—to the company, to investors, to candidates—and be held accountable for progress.
Tie compensation to outcomes. If diversity matters, it should affect bonuses and performance reviews for hiring managers and executives.
Make room for discomfort. Building diverse teams requires examining biases, changing processes, and sometimes having uncomfortable conversations. Leaders must model this willingness.
Invest resources. Diversity isn't free. It requires dedicated recruiters, training, tools, partnerships, and programs. Companies that treat it as a "side project" fail.
The VP of Engineering I mentioned earlier? We rebuilt their entire hiring process. Blind resume review, structured interviews, diverse sourcing channels, clear rubrics. Within 18 months, their engineering team went from 15% to 35% underrepresented groups. Not because they tried harder at the old approach—because they changed the approach entirely.
Diversity theater is comfortable. Actual diversity requires structural change.
References
[^1]: SmithSpektrum diversity hiring data, 90+ companies tracked, 2019-2026. [^2]: Forscher, P.S., et al., "A Meta-Analysis of Procedures to Change Implicit Measures," Journal of Personality and Social Psychology, 2019. [^3]: Bertrand, M. & Mullainathan, S., "Are Emily and Greg More Employable Than Lakisha and Jamal?," American Economic Review, 2004. [^4]: Kapor Center, "Tech Leavers Study," 2017.
Building a diverse engineering team? Contact SmithSpektrum for help designing inclusive hiring processes.
Author: Irvan Smith, Founder & Managing Director at SmithSpektrum