The board now signs for the reporting controls
The Financial Reporting Council's review on 29 September 2026 found reporting quality improving, with substantive enquiry rates down to 28%. The five topics it keeps querying are the ones Provision 29 now asks boards to vouch for.
Quentin Casares · · 6 min read
A falling enquiry rate is not the same thing as a board that could prove its numbers.
The FRC's review covers cases opened in the twelve months to 31 March 2026, at companies with year ends between December 2024 and November 2025. It performed 248 reviews, up from 222 the year before, and 28% of them led to a substantive enquiry letter, down from 37% and from 47% two years ago. Anthony Barrett, the FRC's executive director of supervision, pointed to encouraging signs of progress and added that good reporting is about more than compliance.
The improvement is genuine, and the gap between the largest companies and the rest is closing faster than the headline suggests. Fourteen companies had to refer to the FRC's review in their own annual report, down from eighteen. Most of those changes were reclassifications and presentational corrections. Four of them moved profit.
| What the FRC measured | 2023/24 | 2024/25 | 2025/26 |
|---|---|---|---|
| Reviews performed | 243 | 222 | 248 |
| Reviews leading to a substantive enquiry letter | 47% | 37% | 28% |
| Same measure, FTSE 350 | 28% | 26% | 20% |
| Same measure, other companies | 61% | 44% | 35% |
| Reports required to reference the FRC's review | 26 | 18 | 14 |
Five topics, one problem
The FRC now reports a top five rather than a top ten, after stakeholder feedback, and the five are worth reading as a set rather than as a list of accounting technicalities. Cash flow statements head it at 7% of reviews, with financial instruments at 6%, impairment of assets and fair value measurement at 4% each, and revenue at 3%. Impairment has fallen from 10% in a year. Fair value measurement has risen from 1%, lifted partly by the FRC's own thematic work on investment trusts.
What the five have in common is that none of them is simply recorded. The ledger records transactions. These figures are produced by classification decisions, valuation models and data drawn from treasury systems, contract registers, valuation files and spreadsheets that the accounting system never sees. The FRC says the cash flow problems are mainly about how cash flows are classified within the statement and about inconsistency with disclosures elsewhere in the annual report. That is not an arithmetic failure. It is a lineage failure, found at the point where the same underlying fact is presented twice and does not agree with itself.
| Topic queried | Share of reviews in 2025/26 | Where the figure is actually assembled |
|---|---|---|
| Cash flow statements | 7% | Classification rules applied to ledger movements, reconciled against the narrative and the funding disclosures |
| Financial instruments | 6% | Treasury and hedge records, counterparty and facility data, expected credit loss inputs |
| Impairment of assets | 4% | Forecast models, discount rate assumptions and cash generating unit definitions held outside the ledger |
| Fair value measurement | 4% | Valuation techniques, third-party valuations and unobservable inputs under IFRS 13 |
| Revenue | 3% | Contract terms and performance obligation judgements recorded in sales systems and contract registers |
The FRC's two thematic reviews of 14 October 2026 make the same point from the other direction. The financial instruments review looked at 20 listed corporates under UK-adopted IFRS and found no material issue with how IFRS 9 was applied, while asking for better IFRS 7 disclosure on hedge accounting, expected credit losses and debt factoring. The intangible value drivers review covered 10 reports and asked companies to connect the strategic report more closely to the financial statements. In both cases the accounting was sound and the connective tissue was thin.
Provision 29 changes who carries the evidence
Until now, a disagreement with the FRC about classification was an accounting conversation. Provision 29 of the UK Corporate Governance Code applies for the first time to financial years beginning on or after 1 January 2026, which means boards with a December year end are living inside the first in-scope period now and will publish the first declarations in 2027. It asks the board to declare, on a comply or explain basis, that its material internal controls are effective. The board decides which controls are material. The declaration reaches compliance, operational and reporting controls, including controls over narrative and ESG reporting, and the FRC expects the assessment to rest on evidence from the board's ongoing monitoring and its annual review of the risk and control framework. The Code does not require external assurance, so the evidence is the company's own.
This is the shape banks already know from BCBS 239, arriving in a different regime. A declaration of effectiveness is only answerable if someone can name the Critical Data Elements behind each material reported figure, say who owns them, and produce the lineage from source system to disclosure. Where that record does not exist, the honest route is the explain side of comply or explain, which is a disclosure a board has to be willing to write.
IFRS 18 applies to periods beginning on or after 1 January 2027 and changes how the income statement is presented and disaggregated. Any mapping built to support the first Provision 29 declaration will be rebuilt a year later, which argues for documenting the data path rather than the current presentation.
Oversight has to survive the drafting tools
The second of the FRC's seven expectations for next year is that effective oversight and control is maintained where AI is used in the reporting process. Its research of 8 July 2026, carried out with academics at Lancaster and Loughborough, found generative AI emerging mainly in narrative reporting and still little used in the financial statements, with deployment concentrated on lower-risk, task-specific work. Preparers named trust, data quality and governance controls among the barriers. The FRC also records that complaints to it rose sharply this year and that several appeared to have been drafted using AI.
The FRC's position is unchanged by any of it: companies and their directors remain accountable for the report and accounts. A tool that redrafts a disclosure and quietly carries forward a superseded figure has caused a reporting control failure, not a drafting error, and under Provision 29 that control is one the board has declared effective.
Four questions before the next audit committee
- For each of the five topics the FRC keeps querying, can we name the source system, the owner and the lineage from that source to the published disclosure?
- Which controls has the board decided are material for the Provision 29 declaration, and what evidence from this year's monitoring would support it?
- Where the same figure appears in the financial statements and in the strategic report, what reconciles the two, and who checks it before publication?
- Where AI is used anywhere in drafting or assembling the report, who reviews its output, and would we be able to show that review happened?
The board is not declaring that every number is right. It is declaring that it would know if one were not.
Sources
- FRC, Annual review recognises continued high standards of corporate reporting, 29 September 2026
- FRC, Annual Review of Corporate Reporting 2025/26
- FRC, UK Corporate Governance Code and Provision 29 guidance
- FRC, New thematic reviews provide practical examples to strengthen corporate reporting, 14 October 2026
- FRC, Corporate reporting remains human-led amid growing adoption of artificial intelligence, 8 July 2026
- Burges Salmon, FRC 2026 Annual Review of Corporate Reporting, 21 October 2026
- Baker McKenzie, United Kingdom: FRC Annual Corporate Reporting Review 2025/26, 1 October 2026