Europe rewrote the CSDDD in the name of simplification. Read from the supplier side of the chain, the ledger of gains and losses looks very different from the one Brussels published.
On 26 February 2026, the European Union published Directive 2026/470 in its Official Journal, and the long argument over the omnibus formally ended. The amending text entered into force on 18 March, and the Corporate Sustainability Due Diligence Directive that survives it is a substantially different law from the one adopted in the summer of 2024. I have spent the months since reading the final text the way I read every instrument on this tracker, which is to say from the wrong end of the questionnaire. What follows is not a summary of the changes, because a dozen law firms have published those already. It is an attempt to price the changes for the people who will answer for them.
What was promised, and what arrived
The directive adopted in 2024 promised a phased regime reaching companies with more than a thousand employees and 450 million euro in turnover, a harmonised civil liability route for victims, penalties floored at five percent of worldwide turnover, and an obligation not just to adopt but to put into effect a climate transition plan. Roughly thirteen thousand companies stood inside that perimeter, and every one of them would eventually have needed to know its chain of activities well enough to answer for it.
The directive that entered into force in March applies to companies with more than five thousand employees and 1.5 billion euro in net worldwide turnover, which shrinks the population of directly obligated firms to roughly six thousand. The three application waves collapsed into a single date of 26 July 2029. The harmonised civil liability regime was deleted entirely, leaving injured parties to the mercies of whichever member state's courts they can reach. The penalty floor became a penalty ceiling of three percent. The climate transition plan obligation was removed from the directive altogether, surviving only as a reporting requirement elsewhere. Member states have until 26 July 2028 to transpose, and the Commission owes guidelines and model contractual clauses by 26 July 2027.
Each of those sentences was negotiated by governments, business associations and institutions in Brussels. I am not aware of a single cane farmer, garment worker or mill compliance officer who was asked for a position paper.
The ledger of genuine relief
Honesty requires me to begin with the changes that help, because some genuinely do. The most important is buried in the provisions on prevention and remediation. The original directive required companies, as a last resort, to terminate business relationships where adverse impacts could not be resolved, and every supplier I have ever worked with understood immediately what that clause would become in practice. Termination as a last resort becomes termination as a first reflex the moment a risk register turns amber, because exiting a relationship is always cheaper for the buyer than repairing it. The amended text removes the termination obligation and adds something better: where a company has a reasonable expectation that an enhanced prevention or corrective action plan will succeed, the mere fact of continuing to engage with that supplier cannot itself trigger penalties. That single safe harbour does more for producing regions than any disclosure requirement in the directive, because it makes staying legally safer than leaving.
The second genuine relief is the value chain information cap and the arrival of the VSME voluntary standard as the common language for data requests, with a delegated act due in July 2026. If it works as designed, a supplier below the thresholds should face a bounded, standardised request rather than forty bespoke questionnaires drafted by forty consultancies. The word doing heavy lifting in that sentence is if, and I will return to it.
The third is the shift from entity-by-entity chain mapping to risk-based scoping. Mapping every node of an agricultural supply chain was always a fiction sold as a methodology, and a scoping exercise that concentrates attention where harm is likely is closer to how risk actually lives in a chain.
The ledger of quiet losses
Now turn to the other column entirely. The deletion of harmonised civil liability is the largest single loss for the people this directive claims to protect, and it is a loss precisely because of who bears it. A German buyer now faces liability under German rules, a French buyer under French rules, and a worker in Maharashtra or Sirajganj harmed by the failures of either faces a map of twenty-seven procedural regimes, each with its own standing requirements, limitation periods and costs. Harmonisation was the feature that made the promise of remedy legible from the producing end. Its removal returns remedy to the realm of specialist litigation, which is to say, for most affected people, to the realm of theory. A review clause in 2031 keeps the question technically open, and I intend to still be publishing when it comes due.
The scope reduction cuts deeper than the headline number suggests. Suppliers do not experience the directive through the count of obligated companies but through the procurement behaviour of their specific buyers, and the buyers who dropped out of scope were precisely the mid-sized European traders, brands and processors who sit closest to producing countries in most agricultural chains. The largest multinationals were always going to run due diligence programmes, with or without a directive. The middle of the market was where legal obligation would have changed behaviour that reputation alone never reached, and the middle of the market is what the omnibus excused.
The delay compounds this loss quietly but severely. A supplier weighing an investment in better labour recordkeeping, housing or recruitment channels was previously working against buyer deadlines beginning in 2027. The single application date of 2029, with national transposition trailing into 2028, tells that supplier the market will not price the improvement for years. Compliance investments compete with every other use of scarce capital, and the omnibus just lengthened the payback period on doing the right thing.
The part nobody simplified
Here is the finding that matters most from where I sit, and it will be familiar to anyone who read this publication's first essay. Nothing in the omnibus reduces the information demanded of the supply chain by one questionnaire, because the demand was never really driven by the precise perimeter of the law. It is driven by buyer risk management, and buyer risk management responds to the direction of regulatory travel, not to threshold arithmetic. The six thousand companies still in scope will cascade their obligations downward through contracts, and the seven thousand that escaped will keep cascading theirs anyway, because their banks, insurers and customers now expect it. The VSME standard could discipline this cascade into something bounded and rational, which is why I called it a genuine relief. But standards do not enforce themselves, and the history of this field is a history of voluntary restraint losing to defensive over-collection every single time.
The omnibus simplified the directive for the companies that lobbied for simplification. For the people who answer the questionnaires, it rearranged the furniture, removed the most credible route to remedy, and postponed the incentives while leaving the burden intact. That is not a reason to mourn the original directive, which had real defects, and it is not a reason to dismiss the amended one, which retains real force for the largest chains. It is a reason to keep reading these instruments from the receiving end, because nobody else in the process is going to.