Compassion Benchmark
Special BriefingThematic (event-triggered; candidate for quarterly workforce-governance cadence)June 15, 2026

Layoffs Despite Profits — When a Layoff Becomes a Compassion Failure

A 2026 Fortune 500 restructuring wave is testing a boundary the benchmark is only beginning to price: the difference between a layoff forced by distress and a layoff that protects margin while profits rise. Two cases set the new anchors — Procter & Gamble, downgraded out of the top tier for cutting 7,000 jobs "despite increasing profits," and Oracle, dropped into the Critical band for a 30,000-person cut wrapped in a "sign the release or forfeit your severance" ultimatum. This briefing examines what separates a Boundaries-neutral business decision from a scorable harm.

Scope: Fortune 500 entities tied to the H1-2026 restructuring wave, anchored by the two applied band crossings — Procter & Gamble (Exemplary → Established) and Oracle (Developing → Critical). 22 in-frame F500 entities enumerated below; 2 carry applied layoff-driven downgrades.

Cohort: 22 Fortune 500 entities are in-frame for the H1-2026 restructuring wave the scanners tracked. · 2 carry applied layoff-driven downgrades this cycle: Procter & Gamble (86.1 → 79.0, Exemplary → Established) and Oracle (20.6 → 14.7, Developing → Critical). · P&G's downgrade lands entirely on two dimensions — Boundaries and Integrity (each 4.0 → 3.5) — with the other six untouched. · Oracle's downgrade hits seven of eight dimensions, concentrated in Empathy (5.0), Boundaries (10.0), Accountability and Integrity (12.5 each). · Both crossings emerged from the *same news window* under a shared driver — tariff and cost-reduction pressure — but sit in different bands and different methodology categories.

If you remember one thing

The benchmark now distinguishes distress-driven from profit-protection layoffs. The P&G case turns on a single fact pattern: cuts made while the company is profitable and projecting profits, with the stated driver being tariff cost protection rather than financial distress. That distinction — not the headcount — is what moved the score. A crisis-driven cut of the same size would not have crossed the same threshold.

Key Findings

  1. The benchmark now distinguishes distress-driven from profit-protection layoffs. The P&G case turns on a single fact pattern: cuts made while the company is profitable and projecting profits, with the stated driver being tariff cost protection rather than financial distress. That distinction — not the headcount — is what moved the score. A crisis-driven cut of the same size would not have crossed the same threshold.
  2. A layoff crosses into scorable harm through structure, not size. Oracle's 30,000 cuts are larger than P&G's 7,000, but the decisive scoring element was not scale — it was a coercive-severance architecture: conditioning all severance on signing a legal release that waives the right to sue. This is the first application of a coercive-severance-structure trigger, and it hits Boundaries, Accountability, and Integrity at once.
  3. Profit-period downsizing can break the Exemplary tier on two dimensions alone. P&G's published 86.1 placed it among the benchmark's most-cited positive corporate examples. A single conduct event that weakened only Boundaries and Integrity was enough to end that status — a precedent for any entity near the Established/Exemplary line.
  4. Coercive consent is the sharpest new line. "Sign the release or forfeit severance" converts an ordinary severance negotiation into a consent failure. When a group of laid-off Oracle workers asked for terms matching peer-company layoffs, the request was declined. The benchmark reads that structure as distinct from — and more serious than — the layoff itself.
  5. Layoff conduct is priced only when the harm is verifiable, not when it is merely alleged. Oracle's WARN-Act exposure (counting the 60-day notice period inside severance) was held un-scorable through three cycles because state investigations closed without an adverse ruling. The score moved on the coercive-consent structure, which is documented, not on the unadjudicated WARN theory.
  6. The restructuring wave is a Fortune 500 pattern, not two isolated events. P&G and Oracle surfaced from the same tariff-pressure period that put a broad cohort of large employers — across consumer goods, technology, healthcare, and industrials — into workforce-governance focus. The wave is the context; the two crossings are its first scored expressions.

The field

1,156 entities across the five bands — the full distribution this briefing draws from.

17715%53947%24521%13211%Critical 0–20Developing 20–40Functional 40–60Established 60–80Exemplary 80–100
Source: Compassion Benchmark · CC-BY

1. Frame

A layoff is not, by itself, a compassion failure. Workforces contract; markets shift; a firm in genuine distress that reduces headcount to survive is making a defensible — sometimes unavoidable — business decision, and the benchmark's Boundaries dimension treats it as such. Yet the Fortune 500 restructuring wave of H1 2026 is surfacing a distinct failure mode that the Boundaries and Integrity dimensions are only beginning to price: the treatment of the workforce as an adjustable cost while profits rise, and the use of coercive severance structures that protect the company at the departing worker's expense.

This briefing takes that failure mode as its subject. It asks one central question of the existing record: when does a layoff cross from a Boundaries-neutral business decision into a scorable compassion harm? The June 12 assessment cycle (applied June 14) produced the benchmark's first two answers — Procter & Gamble downgraded out of the Exemplary band, and Oracle dropped into the Critical band — and in doing so established two distinct crossing tests:

  1. The distress-vs-profit-protection test (P&G): cutting while profitable, for margin rather than survival, is evidence against Boundaries (workforce as adjustable cost) and Integrity (external compassion vs internal treatment).
  2. The coercive-consent test (Oracle): conditioning severance on a legal release that waives the right to sue converts a layoff into a consent failure that hits Boundaries, Accountability, and Integrity simultaneously — the first application of a coercive-severance-structure category.

The thesis: the benchmark has begun to price layoffs not by headcount but by two structural features — the financial condition under which they are made, and the consent architecture attached to them — and both anchors are defensible. But one of them (P&G) rests on a primary source whose date sits outside the standard freshness window, and that evidence-timing question must be surfaced honestly rather than buried. Interpreting the applied record is the work here; flagging its one open provenance question is part of that work, not a reason to re-score.


2. The cohort

The in-frame set is the Fortune 500 entities tied to the H1-2026 restructuring wave the scanners tracked across consumer goods, technology, healthcare, industrials, and retail. The table below is recomputed directly from rankings[] in fortune-500.json (composites and dimension vectors confirmed exactly — no drift). Only the two anchors carry applied layoff-driven score movement this cycle; the rest are context entities whose published scores reflect prior assessment, listed to situate the two crossings within the wave.

EntityRankCompositeBandBNDINTLayoff-wave status
Procter & Gamble979.0Established3.53.5APPLIED: 86.1→79.0, Exemplary→Established (7,000 cuts despite profits)
PayPal1877.9Established4.03.5Context — prior workforce-reduction record
Microsoft2665.3Established3.83.0Context — multiple 2025–26 reduction rounds
Citigroup2862.5Established3.53.5Context — restructuring program
UPS3562.5Established4.03.5Context — network/headcount reductions
Dell Technologies4260.9Established3.53.5Context
HP Inc.4460.9Established3.53.5Context
Morgan Stanley6659.4Functional3.53.5Context
Goldman Sachs8650.0Functional3.03.0Context
Starbucks16148.4Functional3.02.5Context — corporate restructuring
Nike17546.9Functional3.03.0Context
Dow17743.8Functional3.02.5Context — plant/headcount cuts
LyondellBasell25135.9Developing2.52.5Context
Walgreens29935.9Developing2.52.5Context — store/corporate cuts
CVS Health32125.6Developing2.02.0Context
3M39420.3Developing2.01.5Context — boundary cluster, just above Critical
Cigna39520.3Developing2.01.5Context — boundary cluster
Oracle41514.7Critical1.41.5APPLIED: 20.6→14.7, Developing→Critical (30,000 cuts + coercive severance)
Boeing42014.1Critical2.01.5Context
Amazon42812.8Critical1.51.6Context — recurring reduction rounds
UnitedHealth Group44510.2Critical1.51.375Context — already Critical (claims-denial, not layoffs)
Meta Platforms4467.8Critical1.51.5Context — three 2026 AI-pivot layoff waves (priced separately)

The two applied crossings are the analytic core. Everything else is the wave around them. Two structural facts stand out immediately:

  • The two crossings are at opposite ends of the index — P&G near the top (rank 9), Oracle near the bottom (rank 415). The same failure mode, layoffs-during-strength, expresses itself at very different absolute scores because it lands on top of very different baselines.
  • Both downgrades route primarily through Boundaries and Integrity — the two dimensions this briefing's theme predicts. P&G moved only BND and INT; Oracle moved seven dimensions, but its deepest are EMP, BND, ACC, INT. The theme is dimensionally legible in the canonical record.

Note the corporate boundary cluster sitting just above the Critical line: 3M (20.3) and Cigna (20.3) are fractions of a point above 20.0, the same pressure zone Oracle occupied (20.6) before this cycle. The Critical line in the Fortune 500 is a thin, crowded band — which is why a single structured-layoff event was enough to push Oracle across it.


3. Two crossing tests — distress-vs-profit, and coercive consent

The two anchors are not the same kind of finding. They establish two separable tests, and keeping them separate is the methodological point.

3a. The distress-vs-profit-protection test (Procter & Gamble)

P&G's case turns on a single distinction the assessment makes explicit: the cuts came "despite increasing profits," with the stated driver being tariff cost protection (a ~$600M pre-tax headwind) plus $1B–$1.6B in restructuring charges — not financial distress. The June 12 digest states the principle directly: "profit-growth-period workforce elimination crosses a Boundaries and Integrity threshold that crisis-driven downsizing would not."

The dimensional record is conservative and surgical. From the assessment:

DimensionPublished (scaled)AppliedΔRationale
BND75.062.5−12.5B1: workforce treated as adjustable cost during profitability
INT75.062.5−12.5I1/I3: tariff pressure absorbed by staff, not by margin; external compassion vs internal treatment gap
AWR/EMP/ACT/EQU/ACC/SYSunchanged0Event does not bear on product safety, awareness, or systemic work

The composite fell from 86.1 (Exemplary) to 79.0 (Established) — a 7.1-point move that crosses a band boundary on the strength of two dimensions alone. The assessor's own note: the published 86.1 "reconstructs exactly under the canonical formula (base 78.1 + integration premium 8.0), so there is no math-hygiene issue. The change is substantive." That matters: the downgrade is evidence-driven, not a reconstruction artifact.

The principle is the contribution. Headcount is not the trigger; financial condition is. A firm losing money that cuts 7,000 jobs to survive would not cross the same line — the Boundaries dimension reads survival-driven contraction as bounded, defensible self-preservation. The harm the benchmark prices is the choice to protect margin with the workforce while the workforce was not the thing under threat.

3b. The coercive-consent test (Oracle)

Oracle's case is structurally different and more serious. The 30,000-person cut (~18% of the global workforce, finalizing by the June 15 departure date) is larger than P&G's, but scale is not what moved the score. The decisive element is a coercive-severance architecture, applied here for the first time:

  • Sign-or-forfeit consent. Per Tech Times: "Employees must sign a release waiving their right to sue in order to receive any benefit at all." Conditioning all severance on a legal release is the consent failure — it is the first application of the COERCIVE-SEVERANCE-STRUCTURE category (June 12).
  • WARN pay absorbed into severance. The 60-day WARN-Act notice period was counted inside the severance calculation rather than paid on top of it — minimizing the actual payout.
  • Forfeited unvested shares. Unvested RSUs were forfeited at termination "regardless of whether those grants were issued as retention incentives or in lieu of salary adjustments" (corroborated across trade and worker-rights coverage).
  • Declined parity request. When a group of laid-off workers petitioned for terms matching peer big-tech layoffs, the request was declined.
  • Margin-over-people framing. Leadership framed the strategy as "choosing the chips" — explicit, public-facing prioritization of capital strategy over workforce.

The coercive-consent structure is what makes this scorable as a Boundaries failure (B5, consent orientation, dropped to 1/5), an Accountability failure (no reparative action, AB5), and an Integrity failure (I1/I3, cost pressure passed to workers). The dimensional move:

DimensionPublished (scaled)AppliedΔ
AWR30.020.0−10.0
EMP5.05.00 (already at floor)
ACT20.015.0−5.0
EQU15.012.5−2.5
BND15.010.0−5.0
ACC20.012.5−7.5
SYS35.030.0−5.0
INT25.012.5−12.5

Composite 20.6 (Developing) → 14.7 (Critical) — a 5.9-point drop across the band boundary. The strongest surviving dimension is SYS (30.0): Oracle has a coherent long-horizon AI/chip strategy. The benchmark's judgment is precise — the strategy is intact; the treatment of the people removed to fund it is what collapsed.


4. Why the same failure mode lands at 79.0 and 14.7

The two anchors are the same theme — workforce-as-adjustable-cost during institutional strength — yet land 64 points apart. This is not an inconsistency; it is the scoring model behaving as designed, and it is worth making explicit.

  • The event lands on the existing profile, not in a vacuum. P&G entered the event from an Exemplary baseline with mature awareness, action, and systemic-sustainability infrastructure across all eight dimensions. The layoff degraded two dimensions and left six high-scoring ones intact, so the composite stayed high (79.0). Oracle entered from a Developing baseline already weak across the board (every dimension below 35.0), so the same kind of event compounded existing weakness and pushed several near-floor dimensions lower at once (14.7).
  • The structure differs, and structure is the multiplier. P&G's event is a financial-condition signal (cut while profitable). Oracle's event is that plus a coercive-consent architecture plus a margin-over-people public framing plus forfeited unvested compensation. More distinct harm axes, more dimensions moved.
  • Within-type severity is coherent at both ends. Within the Fortune 500, 79.0 sits clearly in Established (a strong company with a real but bounded workforce-governance lapse); 14.7 sits clearly in Critical (a company whose worker-facing conduct collapsed). The number reads correctly within the index in both cases. The model is calibrating severity against each entity's own baseline and the number of harm axes, not against a flat layoffs-per-point scale — which is the right behavior for a theme where size is explicitly not the trigger.

The corollary worth stating for publication: a high composite does not immunize against a layoff downgrade, and a low composite does not exhaust the room to fall further. P&G proves the first (Exemplary is not permanent); Oracle proves the second (a Developing company can be pushed into Critical by conduct, not just by accumulated weakness).


5. The pre-adjudication discipline — what was *not* scored, and why

A defining feature of this theme is the benchmark's restraint: layoff conduct is priced only when the harm is documented, not when it is merely alleged. Oracle is the clearest illustration, across three cycles:

  • June 11: A new ruling — WARN-HELD-POST-DEADLINE-BUT-STILL-PRE-ADJUDICATION — established that the June 15 deadline passing and ~30,000 terminations executing does not itself trigger scoring. The Washington and Missouri WARN-Act investigations closed with no adverse ruling and no settlement. "Execution of terminations alone does not constitute an adverse adjudication." Held at 20.6, 0.6 points above Critical.
  • June 12: The band crossing was proposed — but the digest is explicit that it rests on the coercive-consent structure (documented), "distinct from the unresolved WARN-avoidance theory alone." The WARN theory, unadjudicated, did not move the score; the sign-or-forfeit architecture did.
  • June 14 (applied) / June 15 deadline: A DEADLINE-CONFIRMATION-ZERO-DELTA ruling notes the June 15 deadline review produced zero new evidence — the coercive-severance structure was already fully priced at 14.7.

This is the methodologically important boundary for the whole theme: the WARN-Act conduct (counting notice pay inside severance) remains a disputed, unadjudicated compliance interpretation that courts have historically treated as lawful in many circuits. It is not what made Oracle Critical. What made Oracle Critical is the documented, company-confirmed structure of conditioning all benefits on a legal-release signature. The distinction protects the benchmark's independence: it scores verifiable conduct, not litigation theories — and an adverse WARN class-action ruling remains a future conversion trigger that would push Oracle below 14.7, not the basis of the current score.


6. The wave, and the entities on the cusp

The pattern, not the pair. The June 12 digest names the through-line: "tariff and cost-reduction restructuring in H1 2026 is generating a new wave of workforce-governance scoring events in the Fortune 500 index." P&G and Oracle are the first two scored expressions; the cohort in §2 is the wider field the same pressure is acting on. The theme is sector-spanning — consumer goods (P&G), enterprise technology (Oracle, Microsoft, Dell, HP), healthcare (CVS, Cigna, UnitedHealth), industrials (3M, Dow), and retail (Walgreens, Starbucks, Nike).

Adjacent but separately-priced: the AI-pivot layoffs. Meta Platforms (7.8, Critical) carries three 2026 AI-pivot layoff waves (8,000 effective May 20 plus prior rounds), priced under a different methodology family — PLATFORM-SAFETY-TEAM-LAYOFF-AS-DUAL-DIMENSION-EVENT — because the cuts included integrity/safety teams, making them a product-harm signal as well as a workforce signal. Meta is in the wave but on a distinct scoring track; its sector-leading severance (16 weeks + 2 weeks/year + 18 months health) is explicitly credited as having prevented a larger downgrade — the inverse of Oracle, and a useful contrast: severance generosity is a Boundaries-positive that the model reads.

On the cusp. The corporate boundary cluster — 3M (20.3) and Cigna (20.3) — sits where Oracle sat before this cycle: fractions of a point above the Critical line. A single structured-layoff event of the Oracle type would be sufficient to push either across. They are the entities to watch for the next crossing.

Outliers worth naming:

  • Microsoft (65.3, INT 3.0) carries the lowest Integrity score of the upper-band context entities despite a high composite — a profile that, on the P&G precedent, is exposed if a profit-period reduction round is documented.
  • The PayPal/UPS contrast (both BND 4.0) shows context entities that have retained strong Boundaries scores through their reduction histories — the baseline the wave has not yet eroded.

8. Forward view — what to watch

  • The boundary cluster. 3M (20.3) and Cigna (20.3) are the next-likely Critical entrants on workforce-governance grounds. An Oracle-type structured-layoff event at either is the highest-probability next crossing in this theme ().
  • Oracle's downward trigger. The score moved on the coercive-consent structure, not the WARN theory. An adverse ruling in the Washington state WARN class action would add a second, independently-confirmed harm axis and push Oracle below 14.7. This is the single highest-value forward trigger in the theme — it tests whether unadjudicated layoff-conduct theories convert to score on adjudication.
  • P&G scope expansion. The announced 7,000 cuts run over two years. Additional rounds — or above-consensus profit growth reported alongside continued cuts — would warrant a second proposal. The two-year window is a standing monitoring trigger, and the date question should be resolved before any further P&G movement.
  • The Integrity-exposed upper band. Microsoft (INT 3.0 at composite 65.3) and any Established/Exemplary entity with a soft Integrity score is exposed on the P&G precedent: a documented profit-period reduction can break a high composite on two dimensions. The P&G crossing is the precedent the next such case will be measured against.
  • Severance as the positive case. The cleanest signal that the wave is not uniformly a failure would be a large in-window reduction executed with documented above-market transition support that the model credits as a Boundaries-positive (the Meta pattern). Watch for whether any wave entity chooses that path — and whether is resolved in time to score it consistently.

Sources

How to read the scores

The 0–100 scale — five bands

Every entity — state, corporation, AI lab, robotics lab, or city — is scored 0–100 across 8 dimensions and 40 subdimensions. The composite score places the entity in one of five bands:

Critical0–20Foundational compassion practices are absent or documented active harm is present.
Developing20–40Some practices are emerging but remain inconsistent, reactive, or unevenly applied.
Functional40–60Core practices exist and meet a basic bar, with significant gaps remaining.
Established60–80Practices are systematic, documented, and supported by consistent evidence.
Exemplary80–100Practices are independently verified, consistent, and sustained under pressure.

The 8 dimensions

Each dimension is scored 1–5 across 5 subdimensions (40 subdimensions total), then converted to a 0–100 composite. A score of 1.0 on a subdimension represents the minimum anchor; 5.0 is exemplary conduct.

AWRAwarenessDoes this entity reliably detect when others are in pain or need — before they name it?
EMPEmpathyDoes this entity genuinely connect with the inner experience of those it serves?
ACTActionDoes compassionate understanding translate into real, proportional, effective help?
EQUEquityIs care distributed fairly — especially toward those with greatest need and least power?
BNDBoundariesIs helping sustainable, ethical, and autonomy-preserving — not dependency-creating?
ACCAccountabilityDoes this entity own its failures, correct course, and make genuine repair?
SYSSystemic ThinkingDoes compassion extend to root causes and structural change — not only symptom relief?
INTIntegrityIs compassion genuine, consistent, and non-performative — especially when it costs something?

Scores are based on public evidence — government reports, regulatory filings, independent audits, judicial findings, and verifiable third-party records. Entities never pay for inclusion, score changes, or suppression of findings. Full methodology

Continue reading

Companion

June 16, 2026

Allegation, Indictment, Ruling — How the Benchmark Scores Accusations vs Proof

In a single fortnight, OpenAI was hit by a 42-state attorney-general subpoena and its score did not move; Oracle's documented severance terms moved it into the Critical band. That is not inconsistency — it is the discipline that keeps the benchmark citable. This briefing examines six entities to show the exact line the record draws between what is alleged and what is proven, and between conduct an institution chose and conduct a government forced on it.

Read briefing
Companion

June 16, 2026

The Equity Tax — The One Dimension That Drags Almost Everyone Down

The benchmark scores eight dimensions of institutional conduct. One of them — Equity, the fair distribution of care toward those with the greatest need and least power — is the weakest score for nine of every ten entities assessed, from authoritarian states to model corporations. This briefing measures that pattern across all 1,156 entities, shows the exact mechanism by which a single weak equity score caps an otherwise strong profile, and asks what it means that the institutions which get everything else right still fail the most vulnerable.

Read briefing
Companion

June 16, 2026

The Middle of the Scale — What a 50 Actually Means

The benchmark's two foundational briefings spent the extremes: the 23 at the floor and the 64 at the top, together under 9% of the field. But almost every entity a reader looks up — their employer, their city, their country — lives in the vast Developing and Functional middle. This briefing is the on-ramp: what a middling score actually measures, why a balanced 50 and a spiky 50 are not the same thing, and why the "boring" middle is the hardest band to read.

Read briefing
Companion

June 16, 2026

State of Exception — When Governments Codify Impunity

A cluster of governments is not falling to the bottom of the scale through single atrocities. It is legislating its way there — converting emergency powers, "extremist" designations, and election repression into durable, signed-into-law impunity. This briefing tracks that pattern across the Critical-band countries and examines its sharpest case: Bolivia's descent from 28.4 to 6.3 across four scoring cycles, the benchmark's first sequence in which a predicted trigger was named in advance and then realized.

Read briefing
Companion

June 16, 2026

The State of Institutional Compassion — 2026

This is the first comprehensive read on how institutions worldwide recognize, respond to, and reduce suffering. Across seven indexes, 1,156 institutions — every kind, from sovereign states to single-product labs — are scored on one shared 0–100 framework. The headline is sobering and consistent: the modal institution is mediocre, the tails are thin, and almost every institution on Earth, from the worst to the very best, is weakest at the same thing — fairness to those with the least power. This is the state of the field as of mid-2026.

Read briefing
Companion

June 16, 2026

What the Product Is For — Robotics and AI at the Harm Frontier

Sort the 50 robotics labs and 50 AI labs not by rank but by what their core product is *for*, and one gradient appears in both indexes at once: defense, surveillance, and weapons cluster at the floor; healthcare, accessibility, and assistive technology cluster at the ceiling. Compassion Benchmark is the only institution that scores robotics labs at all — there is no comparator. This briefing examines what that gradient is actually measuring, and where conduct and purpose come apart.

Read briefing
Companion

June 15, 2026

AI Governance Under Pressure — What a Shutdown, a Subpoena, and a Union Vote Actually Tell the Benchmark

In a single fortnight, the US government forced Anthropic to pull its two most powerful models, 42 state attorneys general subpoenaed OpenAI, and Google DeepMind's UK staff voted to unionize over military AI. The benchmark scores how institutions recognize and reduce suffering — not how much external pressure they attract. This briefing examines what each of those events does, and does not, say about an AI lab's compassion score.

Read briefing
Companion

June 11, 2026

What Good Looks Like — Exemplars Across Entity Types

The same 0–100 scale that judges the worst also names the best. At the top, 64 entities across states, corporations, AI and robotics labs, and cities reach the Exemplary band. This briefing asks what high compassion actually looks like in the record — what dimension profile produces it, whether it is earned the same way across entity types, and why even the best institutions share a single, universal soft spot.

Read briefing
Companion

June 11, 2026

The Floor and the Critical Band — How the Benchmark Judges the Worst

A single 0–100 scale ranks states, corporations, AI and robotics labs, and cities together. At the bottom, that shared scale meets four entity types that fail in structurally different ways — and reach the bottom by different mechanics. This briefing examines the 176 entities in the Critical band and the 23 at the absolute floor, and asks what the record actually shows about how the worst are judged.

Read briefing

Related daily briefing

June 15, 2026 — daily benchmark

Cite this briefing

Copy-ready citation string for journalism, research, or academic use.

Compassion Benchmark. "Layoffs Despite Profits — When a Layoff Becomes a Compassion Failure." compassionbenchmark.com/updates/special/layoffs-despite-profits-2026-06-15. Accessed [Month Year]. Independent — entities never pay for inclusion, score changes, or suppression of findings.

For methodology, see compassionbenchmark.com/methodology. Data terms: /data-licenses. Press resources: /media.

You just read a Special Briefing.

Weekly score highlights — institutional compassion findings

The week's top score movements and evidence-linked findings across 1,156 entities, delivered every Friday. Daily briefings publish on the site. Free.

No spam. Unsubscribe anytime. Your email is never shared.