Women in Stem: 7 Fixes to Boost Careers and Retention Now

By: Tiago Santana - Founder & CEO, Gray Group International • Serial entrepreneur and growth strategist who has built and scaled multiple companies across technology, media, and consulting. Expert in growth strategist and editorial voice for a global think tank building companies that advance the human experience

Related reading: Inspiring Women Leaders: Empowering Stories and Leadership Insights | Leadership Development for Women: Strategies for Empowering Future Leaders | Women in Leadership Conference: Strategies for Career Advancement

Key takeaways

• Start with a thorough assessment of your specific requirements before choosing a solution.
• Compare multiple options and verify that each meets your documented criteria.
• Avoid over- or under-investing: the right fit balances cost, performance, and long-term value.

In March 2025, Aisha Khan led a 140-person climate software firm in London with GBP18 million in annual revenue. Her engineering team had 42 open roles, time-to-fill had stretched to 67 days, and women held just 19% of technical jobs. Two senior women engineers left within six months, costing the firm well over GBP200,000 in replacement fees, ramp time.

In This Article:

• Key takeaways
• Why are women underrepresented in STEM?
• What barriers hold women back at work?
• 7 fixes leaders can set up now
• How can teams measure real progress?
• What comes next?

Why are women underrepresented in STEM?

In short: Broadly speaking, underrepresentation comes from stacked barriers across education, hiring, daily culture, and advancement.

Broadly speaking, underrepresentation comes from stacked barriers across education, hiring, daily culture, and advancement. Pipeline matters, but so does loss after entry. Many firms celebrate graduate hiring gains while ignoring who leaves by year three or misses stretch assignments by year five.

A common mistake is treating women in STEM as a branding topic. It behaves more like a systems design problem. If inputs are narrow and decision rules stay informal, outputs stay uneven. For Aisha's London firm, the issue was not applicant volume alone. Internal review showed women were less likely to get high-visibility product work tied to promotion cases.

What do the latest STEM numbers show?

UNESCO has reported for years that women account for about one-third of researchers globally, with UNESCO Institute for Statistics data often placing that figure around 33%. In the United States, the National Science Foundation's Science and Engineering Indicators show women earn a large share of science degrees overall, but remain unevenly distributed by field. The U.S. Census Bureau has also found women hold only about 27% of STEM jobs.

Those splits matter more than broad averages. A company can claim strong gender balance if it counts HR or health roles near technical teams. What many decision-makers do not realize is that product risk sits inside function-level gaps. If your AI team is 85% male while your total workforce looks balanced on paper, your reporting hides the real issue.

Which fields have the widest gender gaps?

Computing, engineering, and physics usually show the largest gaps. NCWIT has long tracked this pattern in computing education and careers. Women earn far fewer computing degrees than men in many countries despite strong performance where they do enter.

The practical lesson is simple. Employers face two distinct problems: lower entry shares in some disciplines and weaker retention once hired. That means different fields need different fixes. A life sciences team and an enterprise software team do not draw from the same degree pools, so their recruitment plans should not look the same.

What barriers hold women back at work?

In short: Entry gaps explain only part of the picture.

Entry gaps explain only part of the picture. Many women who do enter STEM hit frictions that men face less often or experience differently. In our experience working with hundreds of organizations, four barriers show up most: informal hiring rules, uneven access to stretch work, manager bias during reviews, and caregiving penalties.

McKinsey and LeanIn.Org's Women in the Workplace research has repeatedly shown that women face a broken rung at the first step up to manager. That pattern matters for STEM because early leadership roles feed principal engineer tracks, research lead jobs, and future executive benches. If promotion slows early, senior representation stays thin for years.

How do hiring systems filter out talent?

Unstructured hiring filters out talent long before an offer stage. Word-of-mouth recruiting tends to replicate existing networks. Vague job descriptions also deter candidates who do not see themselves matching every line item. Research from Hewlett Packard's internal report famously found men applied when they met about 60% of qualifications while women tended to apply only when they met them all.

Aisha's company learned this directly. The firm relied on employee referrals for 46% of engineering hires across twelve months. Referral quality looked strong on speed but weak on diversity because most referrers came from similar prior employers and universities. After switching to scorecards, calibrated interviews, and skills-based screening tasks, female candidate share at final stage rose meaningfully even before offer rates changed.

Why do culture problems raise attrition?

Culture problems raise attrition because daily experience shapes whether people can do their best work over time. ILO research on care work has shown unpaid care burdens still fall more heavily on women worldwide. In most cases that burden collides with workplaces that reward constant presence rather than clear output.

IBM has publicly discussed returnship pathways for experienced professionals re-entering technical work after career breaks through its Tech Re-Entry program launched in 2021 with partners iRelaunch and Path Forward. Programs like this matter because career breaks can unfairly end technical progression despite strong prior performance. Flexibility built into operating design changes who stays long enough to lead.

7 fixes leaders can set up now

In short: Leaders should compare interventions by system effect rather than PR value.

Leaders should compare interventions by system effect rather than PR value. A poster campaign changes almost nothing operationally. A sponsorship program tied to promotion slates can change who gets seen for critical roles within one cycle. The goal is not to lower standards. It is to make standards visible, consistent, and fair.

Our team typically recommends seven moves: widen pipelines beyond elite networks; standardize hiring; publish promotion criteria; train managers on feedback quality; build sponsorship for mid-career women; improve leave and flexibility design; and run pay equity audits annually with action plans attached. These changes work best when leaders treat them as one operating model.

Fix 1: widen STEM recruiting pipelines

Use more channels than referrals and prestige campuses alone. NCWIT research has long argued that computing participation rises when institutions change belonging cues early and consistently. Employers can mirror that by using returnship programs, community colleges where relevant, women's technical networks backed by actual requisitions, and skill-first screening methods.

For Aisha's firm, one overlooked source was experienced analysts from environmental consultancies who already knew climate data workflows but lacked formal software titles. With six months of targeted training plus paired mentoring, that adjacent-talent path filled eight hard roles faster than external searches had done.

Fix 2: standardize hiring and promotion

Publish role rubrics before interviews start. Use consistent questions tied to job outcomes rather than culture fit. For promotion, define what counts as scope expansion, technical depth, cross-team influence, and business impact at each level.

A common mistake is fixing hiring while leaving promotions vague. Opaque advancement erases entry gains fast because people cannot see a fair path forward once inside the company. Clear criteria help managers make better calls and give employees a real map for growth.

Fix 3: build managers who support growth

Managers control assignments, feedback timing, meeting norms, leave tone, and performance narratives. That is why they shape retention more than almost any policy deck does. Train them on evidence-based feedback, sponsorship behavior, interruption control during meetings, and workload planning around leave transitions.

What many decision-makers do not realize is that high-potential labels often spread informally through manager networks before formal reviews begin. Training plus calibration meetings reduce that hidden sorting effect. When managers are better aligned, fewer people are left guessing about what success looks like.

Fix 4: improve flexibility and leave

Design flexibility around outputs instead of face time whenever role requirements allow it. Pair parental leave with re-entry planning, workload coverage rules, and no-penalty performance windows after return periods where needed.

Taken together, these changes signal belonging at moments where many careers stall or break entirely. If flexibility is framed as an exception, it carries stigma. If it is built into the system, more women are able to stay, recover, and move forward.

How can teams measure real progress?

In short: Measure flow metrics across stages rather than just headcount snapshots at year end.

Measure flow metrics across stages rather than just headcount snapshots at year end. Start with applicant share, interview share, offer rate, acceptance rate, first-year attrition, promotion velocity, pay gaps within level, manager span quality scores, leave return rates, and senior technical representation over time.

In our experience, cohort tracking reveals truths dashboards hide elsewhere. If women's entry representation improves but first-promotion rates lag by two years, your problem sits inside management or criteria clarity rather than sourcing volume alone. The point is to track movement, not just totals.

What retention metrics matter most?

The most useful metrics are survival rates by tenure band, manager-level attrition variance, and time-to-promotion by level. Those figures show where leaks happen. NSF data can tell you field supply trends. Your own HRIS tells you whether your systems waste that supply after hire.

Look closely at year two to year five exits. That is where many organizations lose future senior engineers. For Aisha's company, overall turnover looked normal at 14%. Yet turnover among women engineers with three to five years tenure was double the team average. That insight changed priorities fast.

Where do organizations usually get stuck?

Most get stuck between diagnosis and accountability. They collect demographic data but do not tie outcomes to specific leaders, budgets, or timelines. Another common block is fear of legal risk around measurement, even though many lawful approaches exist when designed carefully with counsel.

Schedule reviews quarterly, not annually. If a metric worsens, name an owner within days. Business leaders in London often face extra pressure here because grant makers, public sector buyers, and global partners increasingly ask for evidence, not intent statements.

What comes next?

In short: Better representation in STEM does not come from inspiration alone.

Better representation in STEM does not come from inspiration alone. It comes from redesigning how organizations source, judge, develop, and keep talent over time. For mission-led firms building AI, climate tech, health tools, or advanced products, that work affects both fairness and product quality.

The winning move is to treat inclusion like any other strategic capability build. Set targets. Audit process friction. Fund manager training. Review results publicly inside leadership teams. When the system changes, retention and performance usually improve together.

Ready to take your women in stem strategy further?