The auditor as a signal

Before you read a single number in a company's accounts, a group of people who know those numbers far better than you do have already told you what they think of them — and most readers walk straight past the message. The statutory auditor who signs the accounts, the chief financial officer who prepares them, the independent directors meant to challenge them: what these people do, and when they do it, is a signal about the numbers that is available before you have opened the profit-and-loss statement. An auditor who resigns mid-term and will not say why; a large, cash-rich company that inexplicably uses a tiny, obscure audit firm; a report that arrives with a key audit matter, an emphasis of matter, or an outright qualification; results delayed twice with no clear reason; a CFO who walks out a week before the board meeting. None of these is a number. All of them are readable.

The forensic point is that these governance signals usually move before the numbers do. A company that will disappoint you next year is often, this year, quietly losing the people whose names are attached to its truthfulness — because they can see what is coming and you cannot, and because a name on a document is a liability they can choose to stop carrying. By the time the reported figures turn, the loss is already realised; the resignation is the leading indicator, the disappointing quarter the lagging one. This carries the forensic stance of Module 54 into its most economical form: there, the method was reconciliation — tie every number to another that should agree with it. Here, the tie is between the accounts and the behaviour of the people responsible for them, and you can run it without any accounting at all. illustrative

An audit is a negotiation, not a photograph. The auditor is paid by the company, wants to keep the engagement, and would prefer to sign a clean opinion; the management would prefer the same. So anything short of clean has survived a process in which both sides were pushing toward clean — which is exactly what makes a qualification, or a resignation, so much louder than it looks. An auditor does not casually qualify a client's accounts or walk away from a fee; those are costly, relationship-ending acts, and when they happen anyway it usually means the auditor concluded that signing was the bigger risk. The signal is strong precisely because the incentives ran the other way.

The timing is what makes it forensic rather than a governance nicety. Financial statements are backward-looking and, at the margin, manageable — a determined management can keep the numbers presentable for several quarters. People are harder to manage. An auditor who has seen something they cannot sign off, a CFO who does not want their name on the next set of accounts, an independent director who would rather resign than approve a related-party deal — these are decisions made now, about risk the numbers have not yet shown. That is why governance signals lead: they are the reactions of informed insiders to information you will only receive later, in worse form, as a number. Without this reading you wait for the number; with it, you read the reaction.

The auditor's report is an escalating ladder, and the first move is simply to know the rungs and where the one you were handed sits.

Read the opinion, not just the word 'audited'. A clean (unqualified) opinion says the accounts give a true and fair view. An emphasis of matter still signs clean but points at something — a going-concern uncertainty, a large disputed claim, a note the auditor wants you to read. A key audit matter names the area that required the most judgement, which is a map to where the estimates, and therefore the risks, live. A qualification goes further and states that a specific part of the accounts cannot be relied upon. An adverse opinion says the accounts as a whole do not give a true and fair view, and a disclaimer says the auditor could not even gather enough evidence to form one. Each rung down is the auditor accepting a more costly, more confrontational statement, so each rung down is more informative.

Weigh the auditor against the company. A large, complex, cash-generating business audited by a small, single-office firm is a mismatch worth a question: either the company cannot attract a larger auditor, or it prefers one more dependent on the fee and easier to lean on. The same logic reads the change — a move up to a larger firm is reassuring, a move down to a smaller or obscure one, especially mid-term and unexplained, is the opposite. A persistently tiny audit fee for a big company can mean the audit is thin.

Read resignations, delays and the calendar. The loudest signals are not in the opinion at all. An auditor resignation before term, particularly one whose stated reason is vague or absent, is the strongest single flag in this module. A CFO resigning close to results, an independent director quitting near a contentious decision, results delayed once and then again: these are people declining to attach their names at the moment of maximum exposure. You run this reading on every company, clean ones included — it is only by seeing what a normal, quiet audit history looks like that the abnormal one announces itself.

The loudness of an auditor signal is not a property of the event but of the setting, and this is the inversion that makes it so easy to misread. Tell a reader "the auditor changed" and, without context, they cannot know whether they are looking at a bank obeying a regulator's mandatory rotation rule — pure noise — or a small developer quietly swapping a Big-Four firm for a one-room practice, close to the loudest thing a filing can say. Tell them "the accounts were qualified" and the reach depends entirely on what was qualified.

The stance is constant — read the people around the accounts, weigh the auditor against the company, and ask what forced the event — but the meaning of any given signal has to be relearned for each business. A reader who takes one sector's rule as the rule will fire the alarm on a bank doing exactly what it is told and sleep through the developer's downgrade.

Take a composite mid-cap that has compounded reported profit at 30% for three years, a market darling, the annual report glossy and confident. A reader who stopped at the headline profit would own it happily. The forensic reader reads the people around the numbers first. illustrative

The audit report is not clean. It carries an emphasis of matter drawing attention to ₹310 crore of receivables from customers the company calls "long-standing," and a key audit matter on revenue recognition on multi-year contracts. Neither is a qualification — the auditor still signed — but both point, deliberately, at exactly the two places a growth story of this shape would be manufactured: the receivable that has not been collected and the revenue estimated ahead of billing. Then the calendar: results delayed by three weeks, the second delay in two years. Then the people: the CFO who built the reporting resigned two months before year-end, replaced by an internal promotee, and one of the two genuinely independent directors stepped down citing "other commitments." illustrative

Read one at a time, each has an innocent explanation. Emphasis-of-matter paragraphs are common; CFOs leave for better jobs; directors are busy; results slip. But the forensic reader does not read them one at a time — he reads the cluster, and the cluster points one way: the auditor has flagged the receivable and the revenue estimate, the results are late, and the two people best placed to know have removed their names from what comes next, all in the same year, against a profit line growing far faster than the sector. None of it proves anything is wrong. All of it tells you where to point the number-work of Module 54 — tie that ₹310 crore to cash actually collected, tie the multi-year revenue to billing — and to do it before you believe the 30%. The habit to build is to read the audit report and the governance page first, so the accounts have to survive the people rather than the people being forgotten behind the accounts.

When an auditor resigns or a CFO leaves near results, the forensic analyst does not accuse — he asks the company to put the reason on the record, in checkable specifics. "Your auditor resigned in September, mid-term, and the filing gave no reason beyond 'pre-occupation.' Can you tell us specifically why they left, whether they raised any issue with the audit committee, and what the incoming firm's scope and fee look like versus the last one?" The question names the event and asks for the one thing an innocent explanation can supply and a guilty one cannot: detail.

A good answer gives a checkable cause and stakes itself on something you can hold it to — the firm's global network exited standalone audits below a certain group size and left four peers the same quarter, the letter says exactly that, no matter was raised with the audit committee, the incoming firm is larger, and the fee is up 15%: "judge us on the successor's first opinion next quarter." An evasive answer reassures without disclosing: "we parted on excellent terms," "these things happen," "our new auditors are very reputable," "the balance sheet has never been stronger." It names no cause, points to no letter or minutes, and reaches for adjectives in place of the record. A resignation with an innocent, documented reason is one a company volunteers the document for; the refusal to give the reason is itself the answer. The follow-up that closes it — "Will you release the outgoing auditor's letter in full?" — forces the document into the open, where "we parted on excellent terms" either is or is not what it says. If the warm, content-free reassurance is allowed to stand, the analyst has accepted a feeling in place of a fact and waved through the single loudest signal the company emitted this year.

The signal locates the question; it does not answer it. An auditor resignation is consistent with a fee dispute, a global firm exiting a category of clients, a director's genuine ill-health — and with a company about to be exposed. The letter alone will not tell you which. Treating a governance signal as proof of fraud is the paranoid error in its purest form, and it is wrong for the same reason it was in Module 54: it stops the investigation where the investigation should begin. Nor can a clean report tell you the accounts are honest — some of the most serious failures carried unqualified opinions right up to the collapse, because a determined management can deceive its own auditor. A clean opinion is one input that reconciles, not a reason to skip the reconciliation.

And the signal cannot substitute for base rates. Most auditor changes are routine, most CFO exits are ordinary, most emphasis-of-matter paragraphs point at risks that never crystallise. A reader who treats every departure as sinister generates far more noise than signal, exhausts himself, and eventually stops looking — the same burnout that kills the paranoid number-reader. So the traps are symmetric. One is reading "audited" as a binary stamp and never opening the graded opinion at all, so an emphasis of matter or a qualification passes unread. Another is importing a signal from the wrong sector — crying wolf at a bank rotating on the regulator's cycle, or trusting a small promoter-run company because it "changed auditors just like the banks do." The third is reading the numbers before the people and never revisiting them: because the reported figures are concrete and the resignations feel "soft," the reader anchors on the profit and treats the CFO's exit as colour to explain away. But the governance signal is the leading indicator; reading the number first and the people as a footnote is reading the clock backwards. The fix is sequence — audit report and resignations first, so the people set the questions the numbers then have to answer.