Workload Equity Analysis

Diagnose whether non-promotable (glue) work is distributed fairly across a team, with attention to demographic patterns (gender, seniority, tenure) and role boundaries. This skill combines Tanya Reilly's "glue work" framework with the HBR research on volunteerism bias (women volunteer 48% more often for non-promotable tasks; managers ask women 44% more often) to surface equity risks and recommend concrete rebalancing actions.

Output: a private diagnostic report (~/bettersense-work-reflections/workload-equity/<date>-team-<team>.md) suitable for:

• Manager self-reflection before calibration/promotion cycles
• DEI committee review (anonymized)
• One-on-One discussions about role clarity and expectations
• Process design to make glue work visible and shared

When to apply

Trigger this skill when:

• The manager suspects certain team members (often junior, often women/underrepresented) are over-volunteering for non-promotable work.
• Preparing for promotion Calibration and wanting to check whether glue credit is distributed fairly.
• Noticing that some engineers never get asked to do cross-team coordination, even though it's senior-level work.
• hearing complaints about "some people always take meeting notes" or "the same people always onboard hires."
• The team has high attrition among junior engineers who seem to get stuck doing only support tasks.
• The user wants to proactively audit before problems surface.

Skip when:

• The team is too small (< 4 people) — patterns are obvious without a formal audit.
• The team is in active crisis mode — address the fire first, equity later.
• The user has no visibility into team workload distribution (individual contributor without management scope).

Prerequisites

Requires either:

• The user has already run glue-audit on the team (provides baseline inventory)
• OR the user has sufficient observational data to estimate team member activities (the skill will help gather it)

The skill will interview the user as a proxy observer; it does not directly access individual calendars or task systems.

The interview flow

1. Scope definition

• "Which team are we analyzing? (Team name, size)"
• "Time period: last sprint, last quarter, or a specific project?"
• "Do you want team-wide patterns, or drill into a specific individual's workload?"
• "Anonymity preference: should names be redacted in the final report?"

2. Team roster

List team members with:

• Name (or pseudonym if anonymized)
• Level (junior, mid, senior, staff)
• Gender (user's perception; only if the user knows and is comfortable sharing — needed for bias signal)
• Tenure on team
• Role classification (IC, EM, TPM if applicable)

3. Glue work inventory per person

For each team member (or for the team as a whole if doing aggregate), prompt:

• "What glue tasks has [person] done in the last [period]?"
• Use the glue-audit taxonomy categories as prompts.
• Estimate frequency/time: "Roughly how many hours per week did they spend on each?"
• "Was this task officially part of their role, or extra?"

If the user doesn't know exact numbers, the skill accepts estimates: "high/medium/low" or "many/some/few."

4. Core work check

For comparison, ask:

• "What are each person's core deliverables in this period? (Features shipped, projects completed, code written)"
• This contrast is critical: glue looks bad if the person is also neglecting core work; glue looks like exploitation if core work is strong.

5. Visibility & credit

For each glue item, ask:

• "Who knows this person did this work? (Just the recipient? Their manager? The team?)"
• "Was this work called out in performance reviews, team retros, or promotion packets?"
• "If this work stopped, who would notice?"

6. Volunteering vs. assignment

For tasks that recur:

• "Did [person] volunteer, or were they asked/assigned?"
• Capture narrative examples: "In the April retro, who said 'I'll take the onboarding' vs. 'We need someone to own onboarding — [person], can you?'"

7. Fairness signals

The skill looks for red flags:

• Single-point dependency: One person is the only one who knows X (bus factor 1 on glue).
• Task ghettoization: Certain tasks (note-taking, admin logistics) always fall to the same person(s).
• Credit mismatch: Person does the work, someone else gets visibility/credit.
• Career penalty risk: Person doing glue has slower code output or missed promotion.
• Demographic clustering: Women/junior staff disproportionately doing support/coordination.

8. Optional: one-on-one reflection check

If the user is the manager of these folks, prompt:

• "What have you asked each person to do in the last quarter?"
• "What have you not asked certain people to do, even though it's senior-level work?"
• "Have you explicitly said 'this is not your job' to glue requests landing in someone's Slack?"

Output structure

---
team: [team name]
period: [date range]
analyst: [user name, role]
anonymized: true | false
---

# Workload Equity Audit — [Team]

## Executive summary
- Team size: N
- Glue work distribution: [description]
- Notable patterns: [e.g., "Women do 62% of team glue; men do 38%"]
- Risk level: low / medium / high
- Top 3 redistribution opportunities

## Per-person breakdown

| Person | Level | Gender | Core work (output) | Glue hrs/wk | Glue type | Visibility | Promotion risk |
|---|---|---|---|---|---|---|---|
| Alice | Senior F | F | Shipped auth service | 12 | onboarding, design review | team | Medium (glue not counted) |
| Bob | Senior M | M | Shipped billing pipeline | 2 | occasional PR review | high | Low (code counts) |
| ... | ... | ... | ... | ... | ... | ... | ... |

## Patterns identified

### Pattern 1: [Name, e.g., "Women over-volunteer"]
- **Evidence:** Alice (F, Sr), Carol (F, Mid), Dana (F, Jr) collectively 68% of team glue hours.
- **Possible drivers:** April retro: "We need someone to take onboarding" → Alice immediately said "I'll do it." No men volunteered. HBR study shows this pattern普遍存在.
- **Risk:** These engineers may stall at promotion because glue work crowds out core work and is undervalued.

### Pattern 2: [Name, e.g., "Single-point dependency glue"]
- **Evidence:** Only Alice knows the onboarding process; only Bob knows the deployment rollback procedure (but that's core infra).
- **Risk:** Bus factor; burnout; unfair if Alice leaves, team struggles.

### Pattern 3: [Name, e.g., "Seniors avoid process glue"]
- **Evidence:** Senior men do almost no "process improvement" or "team health" tasks; those fall to junior women.
- **Risk:** Senior work not modeled as ownership; culture work invisible in promotions.

### Pattern 4: [Name, e.g., "Glue not counted in promotions"]
- **Evidence:** Last promotion packet for senior engineer emphasized code output; design review and mentorship mentioned but down-weighted.
- **Risk:** System reinforces code-only path; glue-doers penalized.

## Recommendations

### Immediate (this sprint)
- **Rotate onboarding:** Assign next new hire to Bob (Senior M) with Alice as backup; Alice's time freed → 8 hrs/week for core project
- **Public credit:** Call out design-review contributions in team meeting; link to incident avoidance
- **Task assignment protocol:** For any non-trivial glue task, manager explicitly assigns rather than waiting for volunteers

### Short-term (next quarter)
- **Update team working agreement:** "Glue work is real work; we track it in our project board"
- **Promotion rubric alignment:** Ladder dimension "Team Multiplier" explicitly includes onboarding, design review, mentorship
- **Calendar protection:** Block recurring glue time (e.g., "design review block" Tuesdays 10–12) so it's visible and bounded

### Long-term (6 months)
- **Glue rotation schedule:** Create a shared ownership model; no one owns a glue category exclusively
- **Promotion packet examples:** Collect successful promo cases where glue work was central; share with team
- **Manager bias check:** Before asking someone to take a glue task, ask: "Would I ask a man/white engineer/ senior person to do this?"

**If the user is the glue overload person themselves (self-audit revealed they're the sole owner of 3 categories):**
- Immediate plan: Declare a "glue holiday" for 2 weeks — do not volunteer for any non-core task; let gaps surface
- Schedule a 1:1 with manager: present audit findings, request explicit reassignment of at least 2 glue categories to others
- Set calendar: "Core work blocks" protected; auto-decline meeting invites during those blocks

## Manager guidance section

[Separate section addressed to the user in their manager capacity]
- Psychological safety: this analysis is for system improvement, not individual performance review
- How to present findings to the team without shaming
- How to introduce "glue work is work" into team rituals (retro, planning)
- Rebalancing as career development: giving junior engineers leadership opportunities through structured glue ownership
- Tracking: add glue tasks to project management board like any other work item

## Operating principles

- **Privacy-protective.** Anonymized mode by default if shared; named only if user opts in.
- **Bias-aware but not accusatory.** Surface patterns as "system signal" not "person fault."
- **Action-oriented.** Diagnosis without redistribution plan is just guilt. Always include specific next steps.
- **Link to promotion.** The equity argument is stronger when tied to career impact: "Alice's glue work is senior-level, but our ladder doesn't recognize it — that's a ladder problem or a redistribution problem."
- **Respect core work.** Glue is valuable, but if a person's primary deliverables are suffering, redistribution may need to include offloading, not just sharing.

## Composition with other skills

- **`glue-audit`** — uses its inventory as raw input
- **`promo-case-glue`** — shares artifacts; equity diagnosis supports promotion narrative
- **`manage-glue-workers`** — the redistribution plan becomes the execution instruction for that skill
- **`team-diagnosis`** — this skill is a focused input into that broader health check

## Anti-patterns to flag

- **Naming and shaming.** Never recommend publishing named individual data without explicit opt-in. Use pseudonyms initially.
- **Ignoring core work.** An equitable distribution of glue is meaningless if the team misses all its deliverables. Balance.
- **Assuming intent.** Don't accuse managers of bias; present the pattern and ask if they want to investigate drivers.
- **One-and-done.** Equity is ongoing; recommend quarterly re-audit cadence.
- **Over-counting.** Some glue is genuinely promotable; don't treat all non-code work as "trash." Quality of glue matters (strategic alignment matters more than scheduling meetings).