How much greenwashing is there, really?

Two numbers that run the field

Ask how much greenwashing exists and the answer arrives, almost invariably, as one of two numbers: 42 per cent, or 40 per cent. They anchor regulatory impact assessments, open NGO reports, headline conference decks and justify legislative programmes — usually introduced as what "a recent study" or "research" found. Both numbers are real, and both are almost always miscited. They come from a pair of coordinated website sweeps conducted in November 2020 and announced on the same day — 28 January 2021; they measured online claims across sectors, not packaging; one of them describes a pre-screened sample of already-suspicious claims; and by this article's date they are approaching six years old. The honest answer to this article's title question is uncomfortable for every side of the labeling debate: the field's prevalence evidence is thin, dated, methodologically particular — and the most-cited figures cannot carry the weight routinely placed on them.

This article verifies the two numbers against their primary sources, reconstructs what each sweep actually measured, surveys what newer enforcement evidence does and does not add, and states what can honestly be concluded about greenwashing's extent.

The 42 per cent: the Commission's CPC sweep

The European figure comes from the Commission's press release of 28 January 2021 — "Screening of websites for 'greenwashing': half of green claims lack evidence" — reporting a sweep by the Consumer Protection Cooperation network (European Commission, 2021). The verified mechanics: national authorities screened green online claims across sectors — "garments, cosmetics and household equipment" among them — and examined in detail 344 claims assessed as "seemingly dubious". Of those, "in 42% of cases authorities had reason to believe that the claim may be false or deceptive"; "[i]n more than half of the cases, the trader did not provide sufficient information for consumers to judge the claim's accuracy"; "in 59% of cases the trader had not provided easily accessible evidence to support its claim"; and "[i]n 37% of cases, the claim included vague and general statements such as 'conscious', 'eco-friendly', 'sustainable'" (European Commission, 2021).

Three properties of the number follow from the method, and each is routinely lost in citation. It is conditional: 42 per cent of 344 claims already flagged as dubious — not of green claims generally; the sweep design contains no denominator from which population prevalence could be computed. It is provisional: authorities "had reason to believe" claims "may be" false or deceptive — a screening judgment, not a finding of infringement; none of the 344 had been adjudicated anything at the time of the release. And it is online: websites, not packaging — the sweep's unit was the web claim, and nothing in the release supports transposing its rates onto the on-pack labels this library documents.

The 40 per cent: the CMA/ICPEN sweep

The Anglophone figure comes from the same day's CMA release — "Global sweep finds 40% of firms' green claims could be misleading" (CMA, 2021). The verified mechanics: the International Consumer Protection and Enforcement Network's annual sweep, co-led that year by the CMA and the Netherlands' ACM, was "conducted from 9 – 27 November 2020"; members "swept 1095 websites in total and 495 results have been analysed so far"; of the analysed set, "4 in 10 of these websites appeared to be using tactics that could be considered misleading". The tactics listed lead with vagueness — "[v]ague claims and unclear language including terms such as 'eco' or 'sustainable'" — followed by "[o]wn brand eco logos and labels not associated with an accredited organisation" and "[h]iding or omitting certain information, such as a product's pollution levels" (CMA, 2021).

The same three properties recur, with variations. The figure is impressionistic by design: websites that "appeared to be using tactics that could be considered misleading" — two hedges in one clause. It is partial: 495 of 1,095 swept sites analysed at announcement. And it is again online and cross-sector — "clothes, cosmetics and food" — with packaging nowhere the unit of analysis. One further citation error deserves note: the retrieved release does not say "up to 40%"; sources using that formulation are paraphrasing (CMA, 2021; see verification note).

The third number: 53.3 per cent

Behind the sweeps sits a third figure with better methodology and worse citation practice. The Green Claims Directive's explanatory memorandum reports the Commission's environmental-claims inventory: "The 2020 study found that a considerable share of environmental claims (53.3%) provide vague, misleading or unfounded information about products' environmental characteristics across the EU and across a wide range of product categories", and the accompanying substantiation analysis "found that 40% of claims were unsubstantiated" (European Commission, 2023). These figures derive from a structured inventory rather than an enforcement sweep — a stronger design for prevalence estimation — but they share the vintage problem in aggravated form: 2020 data, published with a 2023 proposal, cited ever since as though contemporary. The 53.3 per cent is, at this article's date, the best available EU-wide prevalence estimate and six years old; the two facts belong in the same sentence whenever the number is used.

What the newer record adds — and does not

The obvious question is whether anything has replaced the 2020 vintage, and the answer, at the verification date, is no — not because nothing has happened, but because what happened is a different kind of evidence.

The post-2021 record is enforcement outcomes, not prevalence measurement. The CPC network's actions — Zalando's commitment to remove sustainability icons (2024), the aviation action ending with twenty-one airlines' commitments (November 2025), the bottled-water commitments clarifying "100% recycled" scope — document that specific practices existed at specific firms and were negotiated away (the enforcement article); they carry no denominator and support no rate. National sweeps and sector studies have accumulated similarly: informative about practice types, silent about prevalence. No coordinated cross-sector sweep with an updated headline percentage was identified in this library's searches — an absence claim, but one consistent with how the 2021 figures continue to dominate citation six years on. The measurement gap is structural: sweeps are enforcement tools that generate publicity numbers as by-products, and no institution in the field owns the task of measuring greenwashing prevalence on a defensible sampling frame at intervals. The Green Claims Directive's verification machinery would incidentally have created such a record; its suspension (the Green Claims Directive article) leaves the gap in place.

For packaging specifically, the gap is doubled. Neither sweep measured on-pack claims; the packaging-specific investigations that exist — the recyclability studies examined in the next article — measure claim accuracy against infrastructure, a different construct from the sweeps' claim-presentation tactics. The honest statement is that no defensible estimate of the prevalence of misleading on-pack environmental claims exists, for any major market, of any vintage. The field legislates, enforces and editorialises atop numbers measuring something adjacent.

Why the numbers persist

The citation dynamics deserve analysis, because they exemplify a failure mode this library documents across the field. The figures persist because they are institutional: regulator-issued, quotable, and attached to the legitimating narrative of the enforcement decade — the EU's blacklist and the UK's code both trace their public justification to the 2021 releases. They persist because they are unfalsified: with no successor measurement, nothing contradicts them. And they persist through vintage laundering: each new document citing "recent research" refreshes the apparent date, until a November 2020 website sweep surfaces in 2026 as current evidence about packaging. The mechanism is identical to the citation-decay cases in this library's third theme — the misattributed consumer-confusion statistics, the "recent" surveys a decade old — and the remedy is identical: date the datum, name the unit, state the denominator.

None of which is an argument that greenwashing is rare. The enforcement record — undertakings, commitments, judgments, settlements across four jurisdictions — establishes beyond argument that materially misleading environmental claims occur at major firms in major sectors. The KLM judgment's fifteen misleading statements, the fashion undertakings' vague-terms inventory, the bottled-water scope corrections: these are adjudicated or negotiated facts, not estimates (the enforcement and comparative-claims articles). What the record cannot say is how much: whether the practices enforcement finds are the visible tail of a pervasive distribution or the bulk of a modest one. Prevalence and existence are different claims, and the field's rhetoric — on both sides — routinely spends the evidence for the second as though it established the first.

What better measurement would look like

Since the gap is methodological, its remedy can be specified methodologically — and stating the design makes clear both why it has not been built and what any prevalence claim would need to be credible.

A defensible prevalence estimate requires four components the sweeps lack. A sampling frame: a defined population of claims — all packaging SKUs in a market's top retail channels, say, or all on-pack environmental representations in a category — from which a random sample can be drawn, replacing the sweeps' enforcement-guided selection. A claim census unit: an operationalised definition of what counts as one claim (a word, a mark, a panel), without which rates are incomparable across studies — the lesson of this library's third theme, where symbol, code and certification mark carry categorically different assertions. An adjudication protocol: pre-registered criteria for classifying a claim as substantiated, unsubstantiated or misleading, applied by assessors blind to the study's hypotheses — the point where the sweeps' "had reason to believe" and "could be considered" hedges would be replaced by reportable inter-rater reliability. And a repeat cadence: the same frame and protocol re-run at intervals, converting a snapshot into a trend — the property that would finally allow the sentence the field keeps writing without evidence: that greenwashing is rising, falling, or responding to the enforcement decade at all.

None of this is exotic; it is the standard toolkit of content-analysis research, applied at market scale. That it has not been done reflects incentives rather than difficulty. Enforcement agencies sweep to find targets, not to estimate parameters — a random sample is precisely what an enforcer does not want. Industry has no interest in funding a number that could only be used against it. NGOs work at the scale of investigations, not census designs. And academic research in this field, as this library's evidence theme documents, has concentrated on comprehension rather than claim prevalence. The measurement gap persists because every institution capable of closing it is optimised for something else — which is itself a finding about the field's governance, and one more entry in the case for the independent measurement infrastructure that the suspended Green Claims Directive's verification records would incidentally have provided.

Until such a design exists, the discipline available to any user of the current figures is citation hygiene: date, unit, denominator. It is a modest standard. Most of the field's prevalence rhetoric, on the record this article has assembled, does not meet it.

The honest answer

Stated for the record, with the qualifications attached. How much greenwashing is there? On the only cross-sector measurements available: in late 2020, among online green claims, European authorities found reason to doubt 42 per cent of a pre-screened dubious sample, a global sweep found four in ten analysed websites using potentially misleading tactics, and a structured EU inventory found 53.3 per cent of claims vague, misleading or unfounded — with 40 per cent unsubstantiated. Those findings are six years old, measure websites rather than packages, and mix screening judgments with prevalence designs. Since then: abundant proof of existence, no measurement of extent. Any sentence that begins "studies show X per cent of green claims are misleading" without a date, a unit and a denominator is — whatever its number — an instance of the practice it describes: a claim whose specification is not provided in clear and prominent terms. This library's own uses of the sweep figures, here and in the companion articles, carry their dates and denominators for exactly that reason — the standard is only worth stating if the stater meets it. Six-year-old numbers can still be cited honestly; what they cannot honestly be is recent — and recency, not accuracy, is the claim their usual deployment makes. The field deserves a better number; until one exists, it owes its readers the dates on the ones it has.

References

CMA (Competition and Markets Authority) (2021) Global sweep finds 40% of firms' green claims could be misleading, 28 January. Available at: Open source (Accessed: 18 August 2026).

European Commission (2021) Screening of websites for 'greenwashing': half of green claims lack evidence, IP/21/269, 28 January. Available at: Open source (Accessed: 18 August 2026).

European Commission (2023) Proposal for a Green Claims Directive, COM(2023) 166 final, explanatory memorandum. Available at: Open source (Accessed: 18 August 2026).

European Commission (2025) Twenty-one European airlines agree to modify their practices regarding environmental claims, IP/25/2608, 6 November. Available at: Open source (Accessed: 18 August 2026).

European Commission (n.d.) Sustainable consumption actions (CPC coordinated actions overview). Available at: Open source (Accessed: 18 August 2026).

Note on sources and verification

All figures and quotations from the two 28 January 2021 releases are verbatim from the primary sources: the Commission's IP/21/269 (retrieved via the Commission's own document endpoint) and the CMA's release as published on gov.uk. The ICPEN sweep's field dates (9–27 November 2020), sample (1,095 swept, 495 analysed), and the "4 in 10" formulation are as stated in the CMA release; the release does not contain the phrase "up to 40%", and the fieldwork dating of the Commission's parallel sweep was not separately captured verbatim — the November 2020 timing is verified for the ICPEN sweep and attributed to the coordinated exercise. The 53.3 per cent and 40 per cent inventory figures are quoted verbatim from the Green Claims Directive explanatory memorandum; the underlying 2020 consultant study was not retrieved as a standalone document. The statement that no updated cross-sector prevalence sweep exists is an absence claim based on this library's searches and the Commission's coordinated-actions page as of the verification date. The characterisation of citation dynamics ("vintage laundering") is this library's analysis. Enforcement outcomes referenced are documented in the companion articles cited, with their own verification notes.

Last verified: 18 August 2026.