Summary

  • Steinhoff's December 2017 disclosure did not begin with a completed verdict. The board said new information concerning accounting irregularities required investigation, delayed audited results and announced the chief executive's immediate resignation. It triggered a collapse in confidence because investors could no longer rely on the timing, completeness or economic substance of previously reported numbers.
  • The later public forensic overview, restated accounts and enforcement records described recurring mechanisms rather than one isolated journal entry: transactions with parties presented as independent, income or asset support without adequate economic substance, circular or manufactured documentation, acquisition and disposal accounting, and balances whose recoverability depended on opaque counterparties. Those sources have different scopes and should not be merged into a single universal finding.
  • Cross-border audit accountability is not solved by assigning every failure to a group auditor or every representation to management. It requires the group audit team to control component scope, contradictory evidence, related-party searches, consolidation entries and unresolved red flags, while the audit committee and supervisory board retain a decision trail showing why challenge was accepted, escalated or overridden.
  • The legal aftermath followed separate tracks. Exchange censures, FSCA administrative orders, Dutch professional discipline, creditor restructurings, civil settlements, a Dutch WHOA confirmation, South African convictions and remaining investigations answer different questions under different standards. A reliable current account preserves those boundaries and records dispositions without turning allegations into verdicts.

The disclosure trigger was a loss of assurance

On 5 December 2017, Steinhoff International Holdings N.V. said that new information had come to light relating to accounting irregularities requiring further investigation. Publication of audited results was delayed, the board approached PwC for an independent investigation and chief executive Markus Jooste resigned with immediate effect. An Amsterdam District Court judgment later reproduced the essential announcement and recorded that Deloitte withdrew consent for use of its audit opinion on the 2016 statements that same day.

The Dutch court's 2020 procedural judgment is useful for the contemporaneous sequence, but its ruling concerned an interlocutory civil issue. It is not a merits verdict on every accounting entry, director or auditor.

The immediate accountability failure was therefore epistemic as well as financial. A listed company had published a consolidated picture that investors, employees, lenders and pension savers treated as decision-grade. The announcement removed assurance from that picture before a replacement account of assets, liabilities and earnings was available. A share-price fall reflected not only revised expectations about future retail trading, but uncertainty about which historical numbers could still be trusted, which entities held value and which obligations were complete.

That distinction matters when reconstructing the event. A trigger disclosure is not the same thing as a forensic conclusion. The announcement properly used investigative language. It did not say that every business was insolvent, every acquisition was fictitious or every director knew the same facts. Nor did an auditor's refusal to complete or continue association with particular statements itself adjudicate fraud. Those actions showed that the available evidence no longer supported ordinary reporting and audit completion on the expected timetable.

The control lesson begins before crisis disclosure. A board should maintain a live register of unresolved accounting judgments capable of changing covenant compliance, acquisition values, reported profit or audit completion. Each issue needs an owner, contrary evidence, quantified ranges, affected entities, legal advice where relevant and a date for escalation. If an external auditor says evidence is limited public evidence, the audit committee should see the underlying request, management response and remaining gap, not a softened summary that the audit is merely taking longer than expected.

Market disclosure should also separate what is known from what remains open. The event, the affected reporting periods, the reason prior assurance cannot be relied upon, immediate governance changes and the next verifiable milestone should be stated independently. A company cannot promise a forensic outcome before the work is done, but it can preserve credibility by versioning facts and correcting earlier statements explicitly. Steinhoff's experience demonstrates how quickly confidence disappears when the first dependable fact is that prior information may not be dependable.

The forensic overview identified a pattern, not one bad number

PwC's investigation was commissioned through lawyers for the boards and took more than a year. In March 2019 the company released an overview, not the complete report. The later Supreme Court of Appeal judgment in the forensic-report access case recounts the commissioning, the public overview and the dispute over legal privilege. In November 2024 the court dismissed the appeal against an order requiring access under South Africa's information-access law. That judgment concerns access to the report and the asserted privilege; it does not itself adopt every forensic statement as a judicial finding against each named person.

The overview described a small group of former executives and other persons involved in transactions that substantially inflated profit and asset values over an extended period. It referred to purportedly independent counterparties, transactions lacking ordinary commercial substance, income generated or supported through arrangements that did not reflect the represented economic relationship, and intercompany or third-party balances that accumulated as the structures continued. The reported pattern reached multiple jurisdictions and reporting periods, which is precisely why a single-entity ledger review would have been inadequate.

The mechanism can be understood as a chain of assurance substitutions. A contract can substitute for proof of an independent counterparty. An invoice can substitute for proof of delivery or enforceability. A valuation can substitute for observable cash generation. A payment routed among group accounts can substitute for settlement by the represented external party. A legal-entity label can substitute for a beneficial-ownership analysis. A component auditor's work can substitute for direct group-team understanding.

Each item may look like evidence when viewed alone, yet the chain fails if the underlying economic event is absent or circular.

This does not mean every unusual transaction is improper. Multinational retailers legitimately use special-purpose entities, intellectual property arrangements, buying groups, vendor rebates, acquisition vehicles and intragroup finance. The accountability test is whether the accounting follows economic substance and whether supposedly external support is independently verified. A counterparty's directors, funding, beneficial ownership, prior dealings, address, advisers and cash sources should be mapped. Side letters and amendments should be searched across local systems.

Cash should be traced beyond the first bank transfer when circularity is a risk.

Forensic work also has limits. It is retrospective, targeted and designed around a mandate. It may use interviews, email review, accounting records and legal analysis unavailable during a routine close, but it does not automatically calculate every claimant's loss or establish criminal intent. Public summaries compress qualifications and may omit material protected for litigation or privacy reasons. A responsible article therefore uses the overview to explain mechanisms while relying on final enforcement decisions and convictions for legal conclusions within their actual scope.

Restatement was a reconstruction of the balance sheet

The reporting repair required more than reversing a headline amount. Previously consolidated statements had to be reassessed across years, entities and transaction families. The JSE's 2020 public censure of Steinhoff as a juristic person established that published financial information for 2016, 2015 and prior periods did not comply with IFRS and the Listings Requirements and was incorrect, false and misleading in material respects. It also addressed two asset disposals and imposed fines totalling R13.5 million. The document says that process concluded against the company as a juristic person; investigations of individuals were then separate.

A credible restatement starts with opening balances. If historical profit inflated an asset or receivable, simply reducing current income does not repair retained earnings, goodwill, deferred tax, non-controlling interests or debt ratios. Acquisitions may need purchase-price allocations revisited. Disposals may require the identity and control of the buyer reassessed. Loans and receivables need existence and recoverability testing. Related-party notes must be rebuilt from beneficial-ownership evidence rather than management labels.

Foreign-exchange translation and consolidation eliminations must follow the corrected legal and economic relationships.

The reconstruction also needs transaction lineage. For every material adjustment, reviewers should be able to move from the restatement note to the affected entity, original journal, contract, invoice, valuation, bank movement, consolidation entry and approving person. The old treatment, corrected treatment and reason for change should remain visible. If evidence was unavailable, that absence should be stated and reflected through impairment, derecognition or uncertainty rather than filled with an unsupported estimate.

The distinction between error, irregularity and adjudicated misconduct must survive the restatement. Financial statements correct balances under reporting standards; they do not decide every person's state of mind. A regulator may later determine that a respondent contravened market law. A prosecutor must prove or obtain a plea to a criminal charge. A civil claimant must establish standing, causation and loss or settle without an admission. Using one vocabulary for all four processes makes the account sound simpler but less accurate.

Restatement quality is measurable. Reconciliations should explain how each corrected period rolls forward. Adjustments should be grouped by mechanism and not hidden within a single net figure. The board should disclose which controls changed, how completeness was tested and what unresolved litigation or tax exposure remains. Auditors should report the scope and evidence behind current opinions without implying assurance over the forensic investigation or future recovery. The objective is not a cleaner-looking balance sheet; it is a traceable bridge from unreliable historical reporting to a supportable present position.

Related parties require evidence beyond legal names

Complex groups can hide economic connection without hiding legal incorporation. A company may be registered in another country, use different directors and transact under a formal contract while still depending on funding, instructions or benefits connected to insiders. Ordinary consolidation systems tend to ask whether an entity is controlled for financial-reporting purposes. A robust accountability system asks a wider set of questions: who proposed the party, who introduced its advisers, who funds it, who can direct it, who receives the residual benefit, and whether its commercial capacity exists independently.

The JSE's 2022 censure of former chief financial officer Ben La Grange provides a concrete transaction-level finding. It described the Steinhoff at Work transaction, a false invoice and supporting documents used to record purported buying-group contributions, with money moved between group accounts to create an appearance of payment. The exchange imposed public censures, fines and a ten-year listed-company office disqualification. That decision is an exchange enforcement outcome; it should not be expanded beyond the conduct, periods and respondent it names.

The transaction shows why bank movement alone is weak evidence. A receipt can confirm that cash arrived, but not that an external customer paid it or that revenue was earned. Reviewers need the originating account, ultimate funding source, agreement, service or goods, counterparty confirmation and subsequent return flow. Treasury teams should flag payments that leave and return within a short window, move through entities sharing authorised signatories or are funded by another group company before an audit confirmation date.

Related-party identification should be continuous rather than an annual questionnaire. Directors and senior managers should update interests when they change. Procurement, legal, tax, treasury and corporate-development systems should share stable counterparty identifiers. Network analysis can identify common addresses, bank accounts, advisers and directors, but software should generate questions, not guilt. Every alert needs human examination and a retained conclusion. False positives are acceptable if resolution is documented; silent gaps are not.

The audit committee must receive more than a list of parties already classified as related under the accounting standard. It should see material unusual counterparties, transactions near reporting dates, nonstandard settlement, management overrides and parties that resisted independent confirmation. If management says a party is independent, the evidence supporting that conclusion belongs in the board record. That process would make it harder for fragmented corporate registries and local knowledge to defeat group-level challenge.

Valuation and consolidation amplified weak transaction evidence

An unsupported transaction can affect much more than revenue. It may create a receivable, increase earnings used in a valuation model, support goodwill, improve covenant ratios and influence an acquisition price. Once embedded in opening balances, later transactions may appear to confirm the earlier value. A group expanding through acquisitions is particularly exposed because management and auditors must assess control, purchase consideration, identifiable assets, goodwill, impairment and post-acquisition performance across many systems.

Steinhoff's restructuring materials provide a later, board-sponsored account of the effort to stabilise the group. The scheme explanatory statement describes the 2019 financial restructuring, the forensic work, litigation-settlement proposal and businesses then held by the group. It is an explanatory document prepared to support a proposed compromise, not an independent valuation or a court finding that every projected recovery would occur.

A valuation control must therefore preserve source hierarchy. Observable third-party cash flows carry different weight from a management forecast; a signed agreement with an opaque party is not equivalent to an arm's-length sale; an internal transfer does not establish external demand. Discount rates and growth assumptions matter, but no model sophistication cures an unsupported starting cash flow. Sensitivities should include the removal of disputed income and the failure of a connected counterparty, not only small percentage changes around management's base case.

Consolidation is another accountability junction. Local ledgers may be internally balanced while group reporting is wrong because entities, eliminations or ownership relationships are misclassified. The group close should enumerate every reporting unit, data owner, currency, accounting framework, component auditor and consolidation adjustment. Manual top-side entries require independent approval and evidence. Changes after component sign-off should return to the component team for confirmation rather than entering through an opaque headquarters process.

Data locality can complicate this work. Privacy, employment and secrecy rules may restrict cross-border transfer of emails or personal data. That does not excuse a group from obtaining sufficient accounting evidence. It requires a designed access model: local secure review, documented search criteria, controlled extracts, audit-team access rights and escalation when law prevents inspection. The board should know which jurisdictions or systems could not be searched and how the limitation affected reporting and audit conclusions.

Group audit challenge cannot be delegated away

The Dutch AFM investigated the audit of Steinhoff's 2015/2016 financial statements and filed a disciplinary complaint against the former external auditor. The AFM's July 2021 account of the decision said the Accountantskamer upheld all parts of its complaint, finding that the audit opinion lacked a sound basis, sufficient appropriate audit evidence had not been obtained and professional scepticism over component work was limited public evidence. A three-month temporary removal was ordered. The AFM also said an appeal could be brought; its announcement was not itself proof that the decision was then final.

The published Accountantskamer decision supplies the formal scope and reasoning. It addressed the individual auditor's professional duties, including group-audit work, contrary indications and the evaluation of a German component auditor. The order stated that the temporary removal would take effect after the decision became irrevocable and an execution order issued. Later public references noted that an appeal had been filed, but the authoritative materials used here do not establish its ultimate disposition. The safe current statement is therefore the first-instance finding and measure, with the appeal boundary preserved.

Group audit responsibility is not the same as performing every local procedure again. It means designing component scope around group risk, communicating instructions, evaluating competence and independence, reviewing significant findings and obtaining enough evidence to support the consolidated opinion. When a component has major unusual transactions, law-enforcement inquiries, management disputes or access limitations, a standard reporting package is not enough. The group team must decide whether to inspect files, perform additional work or modify its conclusion.

Contradictory evidence deserves its own register. A legal letter, whistleblower report, tax inquiry, disputed counterparty, missing confirmation or disagreement between auditors should not disappear into email. The register should state the assertion affected, materiality, procedures performed, unresolved elements and the person accepting residual risk. Closing the audit should require explicit resolution or a reporting consequence. Professional scepticism becomes auditable when the file shows what made the team doubt, what it did and why the final answer survived.

Audit committees must challenge both management and auditor. They should meet component leaders for high-risk regions, understand changes in scope and fee, ask what evidence management could not supply, and review late adjustments and waived differences. Auditor independence is essential, but independence does not transfer the board's responsibility for accurate statements. Equally, management deception does not automatically erase an auditor's duty to respond to red flags. Accountability depends on separating duties while preserving the interfaces where evidence can otherwise be lost.

Exchange and market-conduct enforcement answered different questions

JSE enforcement proceeded against the company and individuals under its Listings Requirements. In January 2023 the exchange censured Jooste over Steinhoff's financial information and the Steinhoff at Work transaction, imposed two maximum fines and disqualified him from office in a listed company for twenty years. He sought reconsideration. The JSE's October 2023 enforcement notice records that the Financial Services Tribunal dismissed that application and that the censure and financial penalties remained binding and enforceable.

That is a final public disposition of the exchange process described in the notice, but not a criminal conviction. The JSE's mandate concerns listed-market requirements and the duties of regulated issuers and officers. Its findings can establish that published information breached those requirements and that specified conduct warranted sanction. They do not decide civil damages for every investor or every offence a prosecutor might allege.

The FSCA followed a different statutory path. Its March 2024 penalty order against Jooste found contraventions related to publication of false, misleading or deceptive statements in annual financial statements and reports and imposed an administrative penalty of R475 million including costs. The order states the evidence, factors and respondent's opportunity to make submissions. It is an administrative enforcement decision under financial-sector law, not a criminal sentence.

The FSCA also dealt separately with former executive Dirk Schreiber. Its enforcement-actions portal lists the March 2024 administrative order and a leniency agreement. Leniency is a legally defined disposition reflecting cooperation and agreed treatment; it does not erase the conduct specified in the documents or establish accusations against other people.

Enforcement chronology should be stored as structured data. For each respondent, a public record should identify authority, legal basis, alleged or found conduct, period, procedural stage, sanction, review or appeal and current status. Names must not be collapsed merely because the same transaction appears in several files. This prevents an early investigation notice from remaining online as if no final decision followed, and prevents an administrative finding from being mislabeled a conviction.

Creditor restructuring preserved operations while reallocating claims

Once confidence and funding contracted, the group faced a liquidity and maturity problem alongside the accounting problem. Lenders needed to know which operating businesses could continue, where cash was trapped and whether enforcement would destroy more value than an orderly restructuring. The initial financial restructuring extended maturities and created time for asset disposals, reporting repair and litigation negotiations. It did not restore the pre-crisis capital structure or guarantee that equity retained value.

The Dutch suspension-of-payments administrators' first public report in English explains how the 2017 disclosure led to financing pressure, frozen intragroup facilities, litigation in several countries and the later restructuring. It is an office-holder report prepared for a legal proceeding and expressly says no rights can be derived from it. Its descriptions and estimates are dated evidence, not universal findings about liability or ultimate distributions.

Creditor accountability requires a legal-entity cash map. Retail subsidiaries may generate operating cash while holding-company creditors have claims elsewhere. Guarantees, security, priority, intercompany balances and local insolvency rules determine who can reach value. A group brand does not create one common estate. Reports should distinguish gross debt, claims filed, claims admitted, disputed claims, secured recoveries, restructuring costs and cash actually distributed.

Restructuring governance must also control conflicts. Directors may owe duties to a company whose stakeholder balance has shifted toward creditors. Lender groups can have different security and maturities. Litigation claimants may prefer a settlement while financial creditors prefer enterprise preservation. Valuations used for voting and class treatment should disclose assumptions, dates and sensitivities. An independent court process can test statutory requirements, but cannot make uncertain future business values certain.

The durable control is scenario evidence before distress. A multinational holding company should know how a reporting qualification, covenant breach or loss of market access moves through guarantees and cash pools. It should model which entities can fund themselves, which require waivers and which services are shared. That information supports an orderly response and reduces the temptation to use optimistic valuations or informal transfers to bridge a crisis.

Civil settlement resolved claims by compromise, not a universal merits judgment

Steinhoff pursued a global settlement through parallel Dutch and South African processes. The settlement combined company contributions with arrangements involving insurers, former directors and Deloitte-related funds, and used claim categories and allocation rules to distribute limited consideration. Different claimant groups had different legal theories, purchase dates and jurisdictions. A compromise provided a coordinated route that fragmented litigation might not have achieved, but it necessarily exchanged uncertain claims for defined settlement rights.

The official Steinhoff Global Settlement case-document archive preserves the composition plan, South African section 155 materials, court orders, allocation plan, claim conditions and public reports. It is the right source for what the instruments said and which documents were filed. It is not neutral proof that every claimant received full compensation, that every released defendant admitted liability or that every valuation in supporting affidavits was correct.

Settlement accountability requires transparent arithmetic. The public should be able to distinguish the nominal value of asserted market-purchase and contractual claims from eligible and admitted claims, settlement consideration, costs, disputed allocations and payments. A claimant who releases a larger asserted loss for a smaller distribution has received a negotiated remedy, not a judicial confirmation that the distribution equals the loss. Conversely, a settlement without admissions is still a real legal disposition and should not be described as if all civil claims remained open.

The settlement site's March 2025 completion notice says claims administration is complete and concluded, with no further reviews, claims or payments to be processed and the claims administrator and dispute committee no longer operational. That is the current status of this settlement mechanism. It does not resolve unrelated proceedings, reopen closed claim windows or establish criminal responsibility.

An effective settlement record should therefore retain claimant-class rules, deadlines, objections, court approvals, funding sources, releases and final administration status. It should also preserve the difference between settlement implementation and corporate survival. The group could settle legacy claims yet remain overleveraged. A civil compromise removes defined litigation uncertainty; it does not create operating value or repair governance by itself.

The WHOA process completed the holding-company restructuring

By 2023 the remaining holding-company debt still required a restructuring solution. Steinhoff proposed a Dutch public WHOA plan under which financial creditors received interests in a new unlisted holding structure while shareholders received a contingent economic interest rather than continuing ordinary ownership on the old terms. Some shareholders opposed confirmation, raising valuation, class and treatment arguments.

The Amsterdam District Court's WHOA confirmation judgment, ECLI:NL:RBAMS:2023:4152, confirmed the plan on 21 June 2023 and rejected the refusal request. The judgment is authoritative for the plan, voting, objections and statutory confirmation analysis before that court. It does not adjudicate every historical accounting irregularity or civil claim, and it should not be described as restoring shareholder value.

This later restructuring and the earlier global settlement solved different problems. The settlement compromised legacy litigation claims. The WHOA plan addressed financial debt, maturities and ownership of the residual group. Mixing the two can create false recovery figures or imply that one court approved all elements of the other process. The legal entities, creditor classes, consideration and release effects need separate tables even when they form part of the same long-running response.

The process also clarifies what governance repair can and cannot mean after a control failure. New boards, policies and audits may improve stewardship, but creditors can still conclude that legacy debt exceeds enterprise value. Transferring economic ownership to creditors is a capital-structure outcome, not evidence that every operational control is effective. Former shareholders' contingent recovery, if any, depends on the defined instrument and future value, not on a restoration of their prior shares.

For future groups, restructuring readiness should be connected to reporting controls. The same entity map used for consolidation should show debt, guarantees, security, tax residence, cash restrictions and insolvency venue. If those records diverge, both the audit and a later rescue become slower and more contestable. Data governance is therefore not an administrative afterthought; it is infrastructure for accurate accounts and fair distress allocation.

Criminal outcomes must be stated person by person

South African criminal enforcement produced concrete outcomes in 2024. Former finance executive Andries Benjamin La Grange entered a plea and sentence agreement on one count of fraud concerning documentation used to inflate Steinhoff at Work results and failure to report fraudulent activities. The joint NPA and Hawks official sentencing statement records a ten-year sentence, five years suspended subject to conditions, and cooperation obligations. That conviction establishes the admitted offence and sentence for La Grange; it is not a conviction of every former officer or every transaction in the forensic overview.

Gerhardus Burger separately pleaded guilty to insider trading and received a wholly suspended five-year sentence and confiscation order. The NPA's 2024/25 annual report records both the Burger and La Grange cases as noteworthy outcomes. Insider trading based on a tip before the price fall is legally and factually distinct from preparing false financial documentation, even though both arose around Steinhoff.

Jooste died in March 2024 after warrants had been obtained and before the contemplated criminal appearance. Death ended the possibility of trying him; it did not convert allegations in an arrest warrant into a conviction or erase final administrative and exchange findings already made under other processes. A criminal account should say plainly that there was no criminal verdict against him. It should also avoid sensational detail not required to explain procedural status.

Other cases require the same restraint. A September 2024 Hawks statement about another Steinhoff-related arrest described allegations, charges and a scheduled court appearance in an insider-trading investigation. An arrest and charge are not guilt. Later official outcomes should replace the pending label when available; until then, presumption of innocence controls.

The correct unit of criminal reporting is the defendant and count, not the corporate scandal. Each record should state charge, plea, verdict, sentence, appeal and forfeiture separately. Evidence from a cooperating offender may support later proceedings but does not decide them in advance. This discipline protects defendants and public understanding while still acknowledging final convictions where they exist.

Continuing proof requires records that outlive the crisis team

Corporate repair is often described through new appointments, policies and committee mandates. Those changes may be necessary, but they are inputs rather than proof. Directors, auditors and regulators need longitudinal evidence showing that the same failure mechanism is no longer passing through the system. A policy requiring related-party disclosure proves little unless declarations reconcile to vendor masters, treasury beneficiaries and corporate registries. A new consolidation tool proves little unless manual entries, rejected submissions and late changes are visible and independently reviewed.

The first useful test is recurrence analysis. Every adjustment and enforcement finding should be translated into a control assertion: counterparty independence, transaction occurrence, cash provenance, valuation support, consolidation completeness, audit evidence or disclosure timeliness. Management should then identify all current transactions sharing those attributes, not merely the old entities or names. Testing should report the population, exceptions, evidence and remediation owner. A zero-exception claim without a defined population is not assurance.

The second test is institutional memory. Investigations and restructurings involve temporary advisers who leave with substantial contextual knowledge. A company must convert that knowledge into controlled records with retention periods, access rights and stable identifiers. Email archives alone are limited public evidence because later reviewers may not know which name variants, entities or dates to search. A decision register should link board papers, legal advice, audit requests, transaction evidence and final accounting treatment while preserving applicable privilege and privacy controls.

Regulators face the same continuity problem. An early notice may remain prominent after reconsideration, sanction or closure. The JSE's current issuer-regulation investigations and enforcement portal places the company, La Grange and Jooste documents in a traceable archive. A reader still needs to open the controlling notice for exact scope, but the index helps prevent the public chronology from ending at the first allegation. Other authorities should publish similarly linked case histories with review status and superseded notices clearly marked.

The third test is challenge quality. Minutes should record the evidence directors requested, alternatives considered and reason a conclusion was accepted. Generic statements that a committee discussed impairment or noted an investigation do not demonstrate scrutiny. For a material valuation, the record should show which cash flows were disputed, what independent data was used and how removing a counterparty changes the result. For a component audit issue, it should show access obtained, extra procedures and the reporting consequence if evidence remained unavailable.

The fourth test is stakeholder usability. Restatement tables, enforcement records and settlement updates are often technically complete but difficult to reconcile. A durable public record should use consistent entity names, dates and transaction identifiers. It should distinguish a euro amount in a forensic overview from a rand penalty, a filed claim from a distribution and a suspended sentence from an acquittal. This is not cosmetic communication. Clear categories reduce the chance that management, media or markets carry an obsolete or inflated number into later decisions.

Finally, proof must survive adverse incentives. Controls that work only while a scandal is under intense scrutiny are fragile. Internal audit should test override by senior executives, unusual quarter-end pressure, rapid acquisitions and disputes with local auditors. Whistleblowers need channels outside the reporting line implicated by a concern. Audit committees should periodically commission independent tests of entity completeness and cash provenance, then publish enough about the method and outcome to make governance claims falsifiable without exposing protected data.

The relevant standard is not whether Steinhoff's exact structures could recur under identical names. It is whether a comparable group could again use fragmented ownership information, unsupported transaction evidence, optimistic valuations and divided audit responsibility to produce a convincing but unreliable consolidated picture. Sustained evidence at those interfaces is the only persuasive answer.

Accountability belongs at the interfaces

Steinhoff's most transferable lesson is not that multinational groups are inherently ungovernable. It is that their interfaces need explicit owners. Related-party declarations must connect to procurement and treasury. Acquisition models must connect to later cash performance. Component ledgers must connect to consolidation. Legal investigations must connect to audit risk. Board challenge must connect to market disclosure. Enforcement notices must connect to current procedural status. Restructuring claim data must connect to actual distributions.

A durable control framework would preserve six linked records. First, an entity and beneficial-ownership graph, including directors, banks and advisers. Second, a transaction-evidence ledger linking contracts, performance, invoices, cash and accounting entries. Third, a valuation register showing source data, assumptions, sensitivities and approvals. Fourth, a group-audit issue log covering components, red flags and evidence gaps. Fifth, an enforcement and litigation chronology that distinguishes allegation, finding, appeal and closure. Sixth, a recovery ledger separating claims, consideration, costs and payments.

Automation can help reconcile names, flag circular cash, identify late journals and compare counterparties across systems. It cannot decide commercial substance or professional scepticism on its own. Models inherit incomplete corporate registries and may mistake coincidence for control. Every automated alert requires a documented human conclusion, and every manual override requires a named approver and retained reason. The aim is not surveillance without judgment; it is to stop jurisdictional and system boundaries from becoming evidence gaps.

Boards should test the framework through adverse cases. Can the company identify all entities exposed to one counterparty within hours? Can it trace a material revenue entry to external performance and ultimate cash? Can the group auditor obtain component evidence despite a dispute? Can directors see whether a valuation survives removal of disputed income? Can market disclosure be issued without waiting for every investigation to finish? Can creditors reconcile legal claims to cash paid?

Accountability is proven when those questions produce records, not assurances. The 2017 trigger, multiyear reconstruction, professional and regulatory findings, civil settlement, creditor transfer and person-specific criminal outcomes form a long sequence of different answers. Keeping them separate is not legalistic caution. It is the method by which a complex cross-border failure becomes understandable, contestable and less likely to be repeated.