The engineering director was frustrated. "We're trying to hire diverse candidates," he said. "But whenever we post a role, the applicants are 90% the same demographic. We interview everyone who applies, but the pipeline is just not diverse. What are we supposed to do—lower our bar?"
No. Lowering the bar is the wrong frame entirely. The problem wasn't that qualified diverse candidates didn't exist. The problem was that they weren't applying—and they weren't applying because of how the company sourced, how it presented itself, and what signals it was sending to the market.
When we examined his process, the issues became clear. Job postings used language that research shows discourages certain candidates. Sourcing was entirely passive—posting on job boards and waiting—rather than actively recruiting from diverse networks. The interview panel was homogeneous, sending an implicit message about who belongs. The company had no visible commitment to diversity; candidates who researched would see nothing that suggested they'd be welcomed.
"You're not fishing in the right pond," I told him. "And the fish you want can see that your pond doesn't have others like them."
This is the fundamental issue with the "pipeline problem" framing: it locates the problem outside the company—there just aren't enough diverse candidates—rather than inside the company—we're not doing what it takes to attract and hire diverse candidates. The former is an excuse; the latter is a solvable problem.
At SmithSpektrum, I've worked with dozens of companies on engineering diversity[^1]. The ones that succeed understand that building diverse teams requires changing how they hire, not just wishing more diverse candidates would appear. The ones that fail keep doing what they've always done and expecting different results.
Why Diversity Matters for Engineering
Before discussing how, it's worth being clear about why.
Diverse teams make better decisions. Research consistently shows that teams with varied perspectives outperform homogeneous teams on complex problems. This isn't about checking boxes; it's about cognitive diversity—having people who think differently, notice different things, question different assumptions.
Homogeneous teams have blindspots. If everyone on your team has similar backgrounds, similar experiences, and similar ways of thinking, you'll collectively miss things that someone with different experiences would catch. This affects product decisions, technical design, and how you understand your users.
Your user base is diverse. If you're building products for a broad audience, having a team that reflects that audience helps you build better products. Engineers who understand diverse users build for diverse users—not because they're checking requirements, but because they naturally consider perspectives that feel foreign to a homogeneous team.
Diverse teams attract more diverse talent. When candidates research your company and see people like themselves, they're more likely to apply and accept offers. Homogeneous teams perpetuate their homogeneity; diverse teams attract more diversity.
The business case is clear. The question is whether companies do the work required to realize it.
The Pipeline Problem Is Usually a Sourcing Problem
When companies say "the pipeline isn't diverse," they usually mean "diverse candidates aren't coming to us." This is a solvable problem.
| Intervention | Impact on Diversity | Common Mistake | What Actually Works |
|---|---|---|---|
| Diverse job boards | Low | Only using this | Combine with process changes |
| Bias training | Very low | Expecting behavior change | Awareness only, not action |
| Blind resume review | Medium | Still biased interviews | Must pair with interview changes |
| Structured interviews | High | Inconsistent application | Calibration and enforcement |
| Diverse interviewers | Medium-high | Token representation | Genuine decision-making power |
Passive sourcing—posting jobs and waiting for applications—draws from a pool that's already skewed. The people who see your job posting and think "this could be for me" are influenced by who they see working at companies like yours, by the language in your posting, by your employer brand. If all of these signal "more of the same," diverse candidates may not even consider applying.
Active sourcing means going to where diverse candidates are. Professional organizations for underrepresented groups in tech. Conferences focused on these communities. Bootcamps that specifically serve diverse populations. Universities with diverse student bodies. Your existing diverse employees' networks. Reaching out rather than waiting means your candidate pool reflects who you actively sought, not just who happened to find you.
Referral programs often perpetuate homogeneity. If your team is homogeneous and you rely on referrals, you'll get candidates who look like your current team—people know people like themselves. Some companies have created targeted referral bonuses specifically for diverse candidates, or have worked to expand the networks from which referrals come.
Job posting language matters more than most companies realize. Research shows that certain words and phrases discourage applications from women and other underrepresented groups: aggressive language like "crushing it" and "dominate," extensive lists of requirements (women tend to apply only when they meet all requirements; men apply when they meet some), and exclusionary jargon. Reviewing postings for inclusive language expands who applies.
Employer branding affects who considers you. What do candidates see when they research your company? If your leadership page shows a homogeneous group, if your blog features only certain voices, if your social media presence doesn't reflect diversity, candidates from underrepresented groups may conclude they wouldn't belong. Visible commitment to diversity—and visible diversity itself—affects who applies.
Process Changes That Matter
Beyond sourcing, the interview process itself affects diverse hiring.
Structured interviews reduce bias. When interviewers have discretion to ask whatever they want and evaluate however they want, bias creeps in—not necessarily malicious bias, but the human tendency to favor people who seem similar to us. Structured interviews—same questions, same evaluation criteria, multiple independent evaluators—reduce this bias and produce more consistent decisions.
Diverse interview panels send important signals. When a candidate from an underrepresented group meets an interview panel of all white men, they receive information about who works there and who succeeds. Including diverse interviewers when possible—without over-burdening your diverse employees—helps candidates see themselves at the company.
Bias training for interviewers helps if done right. Not the one-hour checkbox training that makes people feel good without changing behavior—but ongoing training that helps interviewers recognize their biases and evaluate candidates more fairly. Reviewing interview feedback for bias patterns, discussing calibration regularly, and creating accountability for fair evaluation.
Anonymous resume review can help at the screening stage. Removing names, photos, and other identifying information from initial reviews forces evaluators to focus on qualifications rather than responding to demographic cues. This doesn't eliminate bias—it can creep back in during interviews—but it gets more diverse candidates to the interview stage.
Evaluation criteria defined before seeing candidates prevents post-hoc rationalization. If you decide what "good" looks like before you know who you're evaluating, you can't adjust criteria to favor a preferred candidate. Written criteria, applied consistently, improves fairness.
Retention Matters as Much as Hiring
Hiring diverse candidates into a culture that doesn't support them just creates turnover.
Belonging matters. Diverse employees who feel like outsiders—who don't see themselves represented in leadership, who feel their perspectives aren't valued, who experience microaggressions regularly—leave. The cost of turnover exceeds the cost of creating an inclusive environment.
Sponsorship and mentorship help retention. Underrepresented engineers benefit from sponsors who advocate for them, who create opportunities, who help them navigate the organization. If sponsorship happens informally and homogeneously—people sponsoring people like themselves—it perpetuates inequality. Formal sponsorship programs can help.
Career progression must be equitable. If diverse employees aren't getting promoted at comparable rates to non-diverse employees, you have a systemic problem. Analyze your progression data. If there are disparities, figure out why and fix it.
Inclusive culture requires ongoing work. ERGs (employee resource groups), diversity councils, regular culture surveys, accountability for inclusive behavior—these aren't nice-to-haves. They're how you create an environment where diverse employees want to stay.
The work of retention is different from the work of hiring, but they're connected. A company that hires diverse candidates but churns them creates a reputation that makes future hiring harder. A company that retains and develops diverse employees creates a reputation that makes hiring easier.
What Leaders Need to Do
Diversity doesn't happen by itself. Leadership behavior determines whether it actually improves.
Commitment must be visible and sustained. A one-time announcement or a single training isn't commitment. Sustained attention, regular reporting, and ongoing accountability are commitment. Engineers watch whether leaders actually care or whether diversity is just PR.
Resources must be allocated. Active sourcing costs more than passive sourcing. Training takes time. ERGs need support. Inclusive culture requires investment. If diversity is a stated goal but resources aren't allocated, it's not really a goal.
Accountability must exist. Who's responsible for diversity outcomes? What happens if outcomes don't improve? Without accountability, diversity becomes everyone's concern and no one's responsibility.
Data should inform decisions. What's your diversity at each pipeline stage—applicants, screens, interviews, offers, acceptances? Where are you losing diverse candidates? Is retention different for different groups? Data lets you identify where the problems are.
Personal behavior matters. How leaders conduct themselves—who they listen to, who they promote, how they handle microaggressions—sets the tone. Employees watch leaders. If leaders behave inclusively, that normalizes inclusive behavior; if they don't, stated commitments ring hollow.
Common Pitfalls to Avoid
Certain patterns undermine diversity efforts.
Performative commitment without action creates cynicism. Posting on social media about diversity while doing nothing internally damages trust. Employees know when commitments are genuine and when they're theater.
Burdening diverse employees with diversity work burns people out. Asking your handful of diverse engineers to be on every interview panel, to lead every ERG, to represent diversity in every context—that's exploitation, not inclusion. Spread the work.
Lowering the bar is counterproductive. Hiring candidates who aren't qualified—or who are perceived by others as not qualified—reinforces stereotypes and harms the individuals hired. The goal is to find qualified diverse candidates, not to redefine qualified.
Quick fixes don't work. Diversity is a sustained effort, not a program you implement and complete. Companies that treat it as a project to finish will see temporary improvement and then regression.
Focusing only on hiring ignores the ecosystem. If you hire diverse candidates into a culture that pushes them out, you're wasting effort. Hiring, retention, and culture are interconnected.
Competing on diversity can backfire. If every company is recruiting from the same small pool of diverse senior engineers, bidding wars don't help anyone except those individuals. Growing diversity—through bootcamps, internships, junior hiring—expands the pool rather than redistributing a fixed population.
Measuring Progress
You can't improve what you don't measure.
Pipeline metrics at each stage—applicants, screens, interviews, offers, acceptances—show where diverse candidates fall out. If applications are diverse but interview invitations aren't, your screening may be biased. If offers are diverse but acceptances aren't, something in your offer or closing process is failing.
Demographic data on your current team is the starting point. How diverse is your engineering team? How does this compare to the market, to your goals, to peer companies? Honest assessment of where you are informs where you need to go.
Retention data by demographic shows whether your culture supports diverse employees. If diverse engineers leave at higher rates than non-diverse engineers, you have a retention problem regardless of how well you hire.
Progression data shows whether diverse employees advance. Are they promoted at comparable rates? Are they reaching leadership positions? Disparities here indicate systemic issues.
Engagement and belonging surveys reveal how employees experience the culture. Quantitative data tells you what's happening; qualitative data helps you understand why.
Transparency about metrics creates accountability. Some companies publish diversity data; others share internally. Transparency signals commitment and creates pressure to improve.
The engineering director who thought he had a pipeline problem? His company changed approach.
They started actively sourcing from communities where diverse engineers gathered—conferences, organizations, bootcamps. They revised their job postings with the help of tools that flag exclusionary language. They trained interviewers on bias and structured their interview process. They invested in making their existing diverse employees visible—on the website, at conferences, in blog posts.
Within a year, their applicant pool was meaningfully more diverse. Not because the pipeline had changed, but because they'd started actually accessing it.
"I used to think we had bad luck," he told me. "We didn't have bad luck. We had bad process. Once we changed the process, the 'pipeline problem' solved itself."
References
[^1]: SmithSpektrum diversity advisory, engineering team composition, 2019-2026. [^2]: Harvard Business Review, "Why Diversity Programs Fail," 2016. [^3]: McKinsey, "Diversity Wins: How Inclusion Matters," 2020. [^4]: Project Include, "Diversity and Inclusion in Tech."
Building a more diverse engineering team? Contact SmithSpektrum for talent strategy and inclusive hiring process design.
Author: Irvan Smith, Founder & Managing Director at SmithSpektrum