What is greenhushing? An empirical definition

Twelve of the 54 large European reporters whose externally validated climate target did not change between FY2024 and FY2025 cut their climate disclosure faster than their own workforce disclosure. That pattern is greenhushing in the form that can be measured: climate communication that contracts faster than the same firm's own non-climate disclosure while its externally verified climate position stays unchanged. The term has circulated since 2017 without an agreed definition, and our arithmetic on mandatory Corporate Sustainability Reporting Directive (CSRD) statements puts the prevalence under this one at 22%, well below what the survey headlines imply.

The claim that large companies now say less about climate than they do traces mostly to one consultancy survey, in which South Pole reported a pattern it called "going green, then going dark" in about a quarter of more than 1,400 major companies polled. NewClimate Institute (a climate research nonprofit) read the same material and found three problems: the term has no accepted definition, the survey responses had no independent validation, and little empirical evidence shows the trend exists. A survey records what companies say about their communication plans, and published disclosure has to be measured separately.

The term entered the research literature through a stricter design than the survey wave that popularized it, a 2017 audit study in the Journal of Sustainable Tourism. Font, Elgammal and Lamond audited 31 small rural tourism businesses in the UK's Peak District and found their websites communicated 30% of the sustainability practices the audits confirmed. A documented practice register sat on one side, public communication on the other, and the gap between the two could be counted. A systematic review published in Sustainability this January still lists greenhushing among the least examined themes in the tourism literature, nine years after that coining.

Mandatory CSRD reporting now supplies the audit side of that design at the scale of an entire reporting economy. We rebuilt the same two-sided design across two years of assured statements and a registered of verified targets: a confirmed record of position on one side, published communication on the other.

What makes greenhushing computable in CSRD data

A definition that can be computed has to clear two confounds first: disclosure that shrank in every section during FY2025 for reasons unrelated to climate, and the circularity of judging a firm's climate position from the same document whose language is being scored.

Preparers drafted their FY2025 statements while the Omnibus, the EU measure that would go on to simplify the CSRD and cut mandatory datapoints, was still under negotiation, and many shortened every section while they waited. The directive has since entered into force, in March 2026, with the slimmed reporting standards that implement its cuts still pending, as set out in our 2027 scoping post. A firm that shortened its climate chapter by 8% while shortening every other chapter by 8% was simplifying. The measure has to be relative to the firm's own non-climate disclosure for that reason. We use ESRS S1, the European Sustainability Reporting Standards topic covering a company's own workforce, as the control section. EFRAG, the EU body that drafts those standards, reports that climate (E1) and own workforce (S1) were each material for 99% of reporters. Nearly every firm in the panel therefore carries both chapters, and each firm's workforce section can serve as its own control. Across the panel, climate sections grew 11.8% at the median against 3.0% for the control, so the sections did not move together under simplification pressure.

Scoring a firm's underlying climate commitment from the wording of the same statement whose wording is being measured collapses the construct into itself. That circularity is the second confound, and the fix is a source outside the document. The position side comes from the register of the Science Based Targets initiative (SBTi), the nonprofit that validates corporate emissions targets against a technical protocol, and from nowhere else. A validated target is a third-party technical review and cannot be talked into existence.

That yields a definition you can run as code: greenhushing is climate communication that contracts faster than the same firm's own non-climate disclosure while its externally verified climate position stays unchanged. A firm meeting every part of that test kept its verified target on file and wrote less about climate than it wrote about its own staff, in the same year, in the same document.

A reporting team can run the identical check on itself before filing: compute the change in the climate chapter against a control section, and expect questions if climate contracted faster while a validated target remained unchanged in the register. Every input is free and public. The consequence for assurance providers, investors, and journalists is that the same computation sits within their reach.

How many firms cut their climate disclosure

Across the 195 firms whose FY2024 and FY2025 statements our pipeline could parse and segment, climate chapters grew at the median while control sections barely moved, and the greenhush pattern appears in roughly one in five of the firms whose validated target did not change.

The same medians run against the survey narrative: climate at +11.8%, the control at +3.0%, whole documents at +1.0%. The survey narrative predicts the opposite sign. Firms that reported in both years wrote slightly more words on slightly fewer pages, and denser page layout explains most of that difference.

EFRAG's own like-for-like numbers show that sample composition produced most of the apparent collapse in CSRD reporting volume. Average statement length among firms present in both years fell from 108 pages to 103, while the unmatched year-on-year comparison shows a fall to 95. Firms entering the sample in FY2025 pulled the unmatched average down, and the continuing reporters kept writing at close to their prior length.

Underneath the growing median, 63 firms in the panel did shrink their climate chapter, and the two filters reduce that group sharply. Applying the within-firm control keeps the firms that shrank climate faster than their own workforce section. Applying the SBTi condition then leaves 12 firms out of the 54 whose validated near-term target was in force across both statements, the greenhush cell: the firms meeting every part of the definition.

The exhibit below traces the full funnel from EFRAG's assurance repository, the public list it maintains of filed CSRD statements, down to that 54-firm subpanel.

Our free Regulatory Readiness Assessment sits on the same public rulebook this analysis draws on and returns, for a given company profile, which reporting regimes apply and when. It is free to run and asks for headline company facts and nothing further.

Regulatory Readiness Assessment

How much the 22% figure moves under different rules

Somewhere between 8 and 17 of the 54 target-holding firms fall inside the flagged group, depending on two rules: the margin a firm must exceed over its control section, and the boundary drawn around the climate chapter, both settable in several defensible ways.

Requiring a five-percentage-point margin over the control section cuts the flagged group to 15% of target holders. Loosening the boundary of what counts as the climate chapter lifts the same figure to 31%, and each of those specifications carries its own sampling range on top.

One methodological choice moved the answer more than either threshold: whether the EU Taxonomy annex, a table-heavy disclosure made under a separate EU regulation that happens to sit beside the climate chapter, counts as climate content. Excluding it dropped the greenhush share from 30% to 22% and nearly doubled the measured median growth of climate sections. A construct that swings that far on one boundary decision has to publish its boundary decisions.

We also read the biggest measured contractions by hand, and two of the top four turned out to be extraction artifacts (the pipeline cannot tell a moved table from a deleted chapter). One report interleaved sustainability content through its directors' report, and the software that locates each section read the climate chapter as spanning 88 pages. Another moved its Taxonomy tables beside the climate chapter in one year and away from it in the next. Aggregate medians across a panel this size absorb that noise; a published list of named greenhushers built from the same pipeline would not survive the same hand check.

Two reporting cycles yield a prevalence estimate for one year-over-year window, and no trend can be established from one window. A reporting mandate also sets a minimum amount of climate disclosure a firm must publish whatever its communication strategy. The measure accordingly captures relative de-emphasis inside compliant statements, and full silence stays outside what the data can show. The panel skews to large listed firms with established targets, since those populate both EFRAG's assured list and the SBTi register.

Nothing in the pipeline identifies why any firm wrote less, and the FY2026 filings will face the same limit: legal caution, anticipation of the thinner standards, a change of drafting agency, and deliberate strategic silence all produce identical word counts, incl. among the flagged twelve. I would argue the survey wave measured companies' stated fear of speaking, while this panel measures what they published, and the first can be widespread while the second stays rare. Which of those four forces produced the twelve flagged contractions is the open question.

If you are unsure where to start on your own statement, our Corporate Sustainability Readiness Review, a two-week diagnostic, is the entry point. If you already know what you need, get in touch and we can scope it.

Get In Touch

Seyfi Gasilov is partner for Sustainability Reporting & Governance at Gasilov Group, where he works on CSRD and ESRS compliance, ISSB, internal controls, and supply-chain due diligence. Meet Our Partners →

Frequently Asked Questions (FAQ)

Is greenhushing illegal under the CSRD? 

No; the directive requires disclosure of material information, so a compliant statement can still de-emphasize climate relative to other topics without breaching ESRS. Full omission of a material climate matter would be a compliance failure, which is a separate problem from the relative contraction measured here.

Does the Omnibus change this analysis? 

The Omnibus directive entered into force on March 18, 2026, and the revised ESRS implementing its datapoint cuts are due in a Commission delegated act by September 18, 2026, so FY2026 statements will be drafted against a reduced set of required datapoints. A reduced requirement makes section length more of a drafting choice, which makes within-firm relative comparisons more informative for readers of future statements.