Two dates that are not the same thing
The Third Act Amending the German Act Against Unfair Competition was signed on 12 February 2026 and issued on 19 February 2026 in the Federal Law Gazette (Bundesgesetzblatt, BGBl.) 2026 Part I No. 43. Published is not the same as in force. Article 2 of the act sets 27 September 2026 as the date of entry into force. Until then, the previous rules apply to environmental advertising. One exception has been in force since 19 June 2026: Section 5(6) UWG, which concerns distance marketing of financial services and comes from the transposition of a different directive. Anyone who reads that the new UWG has applied since February reads something wrong. For planning, the difference matters in both directions. You are not behind today, but there is no grace period after the cut-off date either. Entry into force and start of application fall on the same day.
The European background is two deadlines in Article 4 of Directive (EU) 2024/825 of 28 February 2024. By 27 March 2026 the member states had to adopt and publish the necessary measures. They must apply those measures from 27 September 2026. The directive amends Directives 2005/29/EC and 2011/83/EU and aims at better protection against unfair practices and better information for consumers. It carries no official short name. EmpCo is an abbreviation established in legal literature and advisory practice, taken from the English title about empowering consumers for the green transition. One point matters for readers outside Germany: the directive applies EU-wide, but each member state has its own transposition act, and the wording, the numbering and the enforcement route differ from country to country. For a shop that sells to German consumers, the amended UWG is what counts, and this page works to German law.
The Green Claims Directive is sometimes confused with the EmpCo Directive or expected as its sequel. It has not been adopted to this day. Proposal COM(2023) 166 final, procedure 2023/0085(COD), is still listed as an ongoing procedure in the EU primary sources. The European Parliament adopted its first-reading position on 12 March 2024. The first-reading position of the Council is still outstanding. Two things follow. First, there are currently no separate EU-level obligations for prior substantiation and verification of environmental claims. What counts is the UWG as amended through EmpCo. Second, this standstill relieves nobody. It changes nothing in the UWG timetable. Anyone who waits for movement in Brussels before touching the catalogue is waiting for the wrong date.
The four environmental prohibitions that apply without a misleadingness test
New in the Annex to Section 3(3) UWG is number 4a: making a generic environmental claim where the trader cannot demonstrate recognised excellent environmental performance behind it. This is a per-se prohibition. Nobody examines whether the claim misleads in the individual case. The definition is what decides: a claim is generic if the specification is not stated in a clear and prominent way on the same medium. A footnote or a linked sub-page is therefore not enough. The counter-test is recognised excellent environmental performance, and the bar is high: the EU Ecolabel under Regulation (EC) No 66/2010, officially recognised Type I ecolabels under DIN EN ISO 14024, or top environmental performance under other Union law. A self-set standard does not qualify. In practice this means that wording such as environmentally friendly, green, eco or sustainable belongs on the test bench from the cut-off date, if no specification stands next to it on the same medium. Whether a single instance really is a generic environmental claim depends on the context, and finally on a lawyer.
Number 4b prohibits an untrue statement about the scope of an environmental claim: a claim about the whole product or the whole business when it in fact relates only to one aspect or one activity. Typical test patterns are the sustainable online shop where only the shipping is offset, and the environmentally friendly product where only the packaging is recycled. Whether a specific claim falls under this is a legal assessment in the individual case. Number 4c covers offset-based claims: claims based on the offsetting of greenhouse gas emissions, stating that a product has a neutral, reduced or positive effect on the environment in terms of greenhouse gas emissions. In its judgment of 27 June 2024, case I ZR 98/23, the German Federal Court of Justice (BGH) held that an ambiguous environmental term such as carbon neutral must as a rule be explained in the advertisement itself, and that a link or QR code is not enough. It did not have to decide whether such an explanation is sufficient for offset-based claims. From 27 September 2026 that no longer matters. Number 4c applies regardless of the explanation.
Number 2a covers labels: displaying a sustainability label that is neither based on a certification scheme nor established by public authorities. The act defines a certification scheme as a third-party verification system with publicly available conditions, non-discriminatory access, defined procedures for infringements and objective monitoring by a third party. A trader who awards a label to itself does not meet this. On top of that comes the breadth of the term environmental claim. It covers text, images, graphic elements and symbols, expressly including labels, brand names, company names and product names. A green leaf icon in the grid view can already be an environmental claim. So can a product name containing Eco. This is exactly why reading the description texts is not enough. The check has to reach names, attributes and image elements as well.
Alongside the four environmental prohibitions there are further changes that matter in practice. Under the new Section 5(3) no. 4 UWG, an environmental claim about future environmental performance is misleading if it is not based on clear, objective, publicly available and verifiable commitments set out in a detailed and realistic implementation plan. That plan must name measurable and time-bound targets and the means required, and it must be reviewed regularly by an independent external expert whose findings are made available to consumers. Under the new Section 5(3) no. 3 UWG, it is misleading to advertise benefits that are irrelevant and do not follow from a feature of the product, the service or the business. Here the misleadingness test still applies. New Annex number 10a prohibits presenting statutory product requirements as a special feature of your own offer, again without a misleadingness test. Annex number 23d covers misleading statements on software updates, durability and repairability. Section 5b(3a) UWG adds information duties for services that compare products by environmental or social features.
How the pre-audit works: three sources, all fields, all variants
The pre-audit rests on three sources that correct each other. First, an automated scan of the publicly reachable shop pages. It shows what a reader actually sees, including category texts and landing pages. Second, a full technical analysis of the product data export across all data fields. It shows what sits in the source system, including places where nothing is output at the moment. Third, manual review of selected hits, cross-checked against the manufacturer information. It separates the real test candidates from the accidental hits. Without the third source the result would be a hit list nobody can work with. Without the second, half of the data would stay invisible. The method is built for large catalogues with deep variant structures, and it has been used on a shop with around ten thousand product records. We do not publish client details, not even when they would read well as a reference.
The check goes beyond the visible product texts. It also covers meta description and meta title, keywords, the short summary, properties, technical data and FAQ fields — exactly the places where wording stays untouched for years, because it does not stand out in the layout. Variants matter just as much, across every level. If a claim is corrected only on the parent product, it stays in the variants and reappears with the next export. This is why the analysis runs over the complete data set and not over a sample. There is a practical side effect. You get a reliable overview of which field holds which text, and which import it came from. That is useful even when part of the hits turns out to be harmless in the end.
We sort the risk patterns by priority. At the top are generic environmental claims without a specification on the same medium. Then percentage savings claims without a visible basis for comparison. Then whole-product claims where the effect in fact depends on a single component or on the packaging. Then absolute promises of effect, that is wording such as prevents, guarantees or completely, which leaves no room and is therefore hard to substantiate. Then labels without a demonstrable certification scheme. Then durability and repairability claims. And last, but not to be ignored, the invisible meta and keyword fields that a visual check in the shop systematically misses. This order is a working order, not an assessment of the legal position. It says where a clean-up starts when time is limited. Which class is most strongly represented in your catalogue depends on where your product texts come from.
What a hit means, and what it does not
An automated term hit is not a legal infringement. It is a test candidate, nothing more. Green can describe a colour. Sustainable can refer to something quite different from the environment in a running text. A word fragment can match by pure chance. In the other direction, a scan misses wording that carries an environmental claim without using any of the search terms. Both are the reason why the manual review belongs to the method and is not sold as an extra. In the report we therefore do not write what is prohibited. We write what has to be examined, and why. A hit list that marks every find as an infringement creates work in the wrong place, and trust in the wrong place as well. Next to every hit we state which field it was found in and why it belongs on the list.
The scope covered has limits, and we name them in advance. Pages that are not indexed, dynamically loaded content, text inside images, PDF documents, banners, structured data and pages that are unreachable during the run lie outside the automated scan. The product data export compensates for part of this, because it maps the source system completely. But it only knows what sits in the source system, not what a theme or an extension outputs on top. Anyone who promises complete coverage promises something an automated check cannot deliver. We would rather have you know which part of your site was examined and which part was not, than have you rely on a number that blurs that distinction.
And here our work stops. The pre-audit is not legal advice and not a legal opinion. Sharpness Solutions GmbH is a digital agency, not a law firm. We tell you which wording sits in which fields, which risk pattern it matches, and in which order a clean-up makes sense. Whether a specific claim remains permissible in the individual case is a legal assessment, and that requires a lawyer. For material claims we expressly recommend that review before final approval. The pre-audit serves internal preparation, priority setting and technical clean-up. It does not replace the legal decision. It prepares it, because the law firm no longer has to search and can assess instead. We give no assurance of legal certainty and no assurance against warning letters. That would lie outside what a digital agency can judge.
Correction standard, action plan and evidence file
The correction standard is simpler than it sounds. Blanket terms are replaced by concrete properties. Instead of environmentally friendly, the text states the material share, the way the product works, or a measurable figure. Savings claims are tied to their conditions and to a defined baseline, instead of standing free in the room. Percentages get what makes them verifiable: comparison product, conditions, period, method and source. Absolute wording is rewritten as a possibility or a reduction, because prevents is a promise and reduces under the stated conditions is a description. The wording does not become softer in this process. It becomes more precise. There is a side effect for the sales texts. Concrete statements read in a more factual way, they are also easier to substantiate, and they can be used as attributes in filters and in search, which is not true of adjectives.
The action plan runs in phases. First, unsubstantiated claims are blocked or replaced, starting with the highest risk class. In parallel, requests go out to the manufacturers: claim-specific evidence, test reports and details of the scope, so that it is clear which articles a document actually covers. Then the source system fields are cleaned up systematically, including the meta data and all variants, so that the correction survives the next import. After that an approval matrix is introduced: claim, approved wording, comparison, conditions, source, validity and approval responsibility in one table that somebody maintains. Once the changes are live, a full scan runs again, because a clean-up without a follow-up check stays a declaration of intent. And finally, the review of supplier texts and certificates is built permanently into your product data process, so that the next catalogue import does not bring everything back.
What belongs in the evidence file can be set out as a minimum list. First, the exact approved wording together with the articles and variants concerned — not the topic, the sentence. Second, the specific comparison product or a clearly defined baseline. Third, measurement method, test conditions, configuration, period and geographic reference, as far as they are relevant to the claim. Fourth, the manufacturer document or an independent test report, each with version and date. Fifth, for labels, the scope together with the expiry or review date. Sixth, approval date, responsible unit and the date for the next review. This file is not an end in itself and not a folder for the shelf. It is the difference between a claim you can explain in a few minutes when it matters, and a claim whose origin nobody in the company remembers. And it is the basis on which a law firm can assess instead of research.