Why Data Alone Won't Fix Gender Inequality at Work

Every firm I've walked into over the past year has a dashboard. Headcount by gender, attrition rates, hiring ratios — the numbers are there, colour-coded and quarterly. And yet, in factory after factory, the share of women on the floor hasn't moved in three years.

This is the central paradox of the gender diversity space right now: we have more data than ever, and less progress than we'd like.

The Measurement Trap

There's a quiet assumption embedded in most corporate diversity programmes — that if you can measure a gap, you're halfway to closing it. It's an appealing idea. It borrows the logic of lean manufacturing: identify the defect, trace the root cause, eliminate it.

But gender inequality isn't a defect in a production line. It's a feature of how organisations were designed — who they were built for, what behaviour they reward, and whose time they treat as flexible.

"What gets measured gets managed — but not everything that matters can be measured, and not everything that can be measured matters."
— Attributed (loosely) to Peter Drucker

When firms set a KPI of 30% women in middle management by 2026, they often achieve the number without changing the underlying conditions. Women are promoted into roles with high visibility but low authority. Or the denominator shrinks — men leave, the ratio improves, nothing structurally changes.

What the RCT Evidence Says

Over the past eighteen months, our team at Good Business Lab has run randomised controlled trials across factories in the automotive sector — testing specific interventions on hiring, retention, and progression.

A few things that moved the needle:

Intervention Effect
Structured interviews with gender-blind scoring +12% women shortlisted
Peer-support groups for first-year women workers 23% reduction in attrition
Manager training on flexible shift requests +8pp satisfaction scores
Revised crèche policy with awareness campaign Statistically insignificant alone

The last row is instructive. Firms love announcing crèche benefits. They're visible, they photograph well, they go in the ESG report. But without a parallel campaign that normalises using the crèche — that makes it socially acceptable for a woman to leave the floor at a specific time — the policy sits unused.

Infrastructure without culture is furniture.

The 12 KPIs We Actually Track

When we built our AI self-assessment tool for auto firms, we resisted the temptation to track everything. After months of iteration, we landed on 12 indicators across four categories:

1. Representation

  • Share of women in total headcount
  • Share of women in supervisory and above roles
  • New hire gender ratio (last 12 months)

2. Retention & Progression

  • Gender gap in 12-month attrition
  • Internal promotion rate by gender
  • Time-to-promotion gap

3. Working Conditions

  • Access to gender-segregated facilities
  • Flexible work policy utilisation by gender
  • Grievance reporting rates

4. Culture & Inclusion

  • Manager awareness scores (from training assessments)
  • Psychological safety index (from anonymous pulse surveys)
  • Peer allyship behaviours (observed and self-reported)

No single number tells the story. The value is in the pattern — which quadrant is strong, which is hollow, and where the leverage is.

What Actually Moves the Needle

After two years in this space, the interventions that consistently work share three properties:

  1. They change the default. Opt-out policies outperform opt-in ones. If women must actively request a flexible shift, most won't. If flexibility is the default and rigidity requires a request, behaviour shifts.

  2. They involve men. Gender diversity programmes that treat men as bystanders fail. The firms where progress is fastest are the ones where male supervisors are held accountable — where their performance review includes team retention and inclusion scores.

  3. They are boring. The high-impact interventions are rarely the ones that make the annual report. Structured interview scorecards. Consistent enforcement of the anti-harassment policy. A supervisor who follows up when someone doesn't show up for three days. Unglamorous, relentless, effective.

A Note on Patience

This work is slow. The firms I've seen make the most progress started five years ago, failed quietly at several things, and kept going. The ones who haven't moved are often the ones who launched a splashy initiative in 2022 and declared victory.

If you're building a gender diversity programme, the most important question isn't what should we measure — it's what are we willing to keep doing when it's no longer new?

Shreya Ganguly works on gender diversity in India's manufacturing sector at Good Business Lab. Views are her own.