Scuttlebutt

The filing tells you receivables grew far faster than sales. The management call tells you demand has never been stronger. Both can be true at once, and the accounts alone cannot tell you which story wins — a receivable running ahead of sales is either a large genuine order awaiting collection or product pushed into a channel that cannot absorb it, and the P&L reads identically in both cases. So you close the annual report, and you go and look. You call a distributor who actually stocks the product. You visit the store and watch whether anyone buys it. You find someone who left the company last year and ask what really broke. This legwork — checking a company against the real world beyond its own paperwork — is what Phil Fisher named scuttlebutt, and it is the subject of this module. illustrative

The word matters, because it warns you what the method is not. Scuttlebutt is not a tip, and it is not a shortcut past the accounts. Its only honest use is triangulation: you build a thesis from the statements — this is a channel-stuffing risk, this margin looks too good, this order book may be padded — and then you go into the world to confirm or break that thesis with independent, observable evidence. The direction runs from the accounts outward, not from a rumour inward. A reader who starts with a whisper and goes looking for a company to attach it to is not doing scuttlebutt; he is laundering a tip. A reader who starts with a reconciled question from the filings and tests it against the shelf, the dealer and the ex-employee is doing exactly what Fisher meant.

This module sits in Part Six because comparison and relative value are only as good as the facts you feed them, and some of the most decisive facts are not in any filing — they are on the ground, in the channel, in the turnover of the people who worked there. But scuttlebutt carries two hazards that make it as easy to fool yourself with as to inform yourself: the human pull to collect only the evidence that agrees with you, and a hard legal line — the point at which "research" becomes trading on material non-public information, which is a criminal offence, not a clever edge. Learn the method and you gain a real check on the numbers; ignore its two hazards and it becomes the most confident way to be wrong.

Everything earlier in the guide taught you to read what the company chose to publish — the statements, the notes, the ratios, the management's own words. This module exists because the published record, however carefully read, is a self-report: the company selected what to disclose and framed how to disclose it, and there are questions the report raises but cannot answer about itself. When the accounts throw up a gap — profit that has not become cash, a receivable outrunning sales, a margin that no competitor can match — the reconciliation locates the question, but the answer often lives outside the paperwork, in the physical reality the company operates in. Scuttlebutt is the discipline of going to get that answer honestly.

The founding idea is Fisher's, from Common Stocks and Uncommon Profits: that a diligent investor can learn an enormous amount about a company by asking the people around it — customers, competitors, suppliers, former employees, dealers — none of whom has the company's incentive to flatter it, and each of whom sees one true facet the company would rather blur. A competitor will tell you, often with relish, where a rival is weak. A supplier knows whether orders are accelerating or being quietly cancelled. A distributor knows whether the product actually sells through or merely ships in. An ex-employee knows why the last three finance heads left. No single one of these is decisive, and none is a substitute for the accounts; together, read against a thesis, they either corroborate the story the numbers tell or contradict it, and a contradiction is worth more than a page of reassurance.

Without this stance, two failures follow. The first is the reader who trusts the self-report entirely — who takes "demand is strong" from the concall as a fact rather than a claim, and never discovers that the strength is inventory sitting in the channel, not sales to end customers. The second, subtler and more common among those who do legwork, is the reader who uses scuttlebutt not to test his thesis but to decorate it — who calls three sources, keeps the one that agrees, and calls the result "channel checks." The method's value comes entirely from independence and from a willingness to be broken; a scuttlebutt exercise that cannot come back negative is not research, it is a ritual of self-persuasion. This module installs the honest version and marks, in bright paint, the legal line that the dishonest version eventually crosses.

Scuttlebutt has a shape, and the shape is what keeps it honest: a thesis from the accounts in the centre, independent observable checks feeding in, and a verdict — confirmed or broken — coming out.

Start from a thesis, not a blank curiosity. The single move that separates research from gossip is direction. You do not ring a dealer to ask "what do you think of the company?" — you ring having already decided, from the accounts, what you need to know: does the product sell through, or is it being stuffed into the channel to book the sale? The reconciled question from the filings — the receivable that outran sales, the inventory that keeps building, the margin no peer earns — is the thing scuttlebutt is sent to test. This is why the module depends on everything before it: without a thesis, scuttlebutt has nothing to confirm or break, and it degenerates into collecting impressions that flatter whatever you already hoped.

Go to the sources with no incentive to flatter. The company's own channels — investor relations, the annual report, the roadshow — are self-reports. The value of scuttlebutt is that it reaches people whose interests are not aligned with the share price. A dealer or distributor sees whether stock moves off the shelf or piles up in the back — the truest read on genuine demand for a consumer or auto business. A supplier or sub-contractor sees order flow accelerating or being cancelled before it reaches revenue. An ex-employee — found through public professional networks, spoken to about their own experience, never about anything confidential — knows why people left and what the culture hides. A customer, or a store check you do yourself, tests whether the buyer agrees with the seller's story. Physical channel-inventory signals — how much stock sits with distributors, how long it has been there — turn "demand is strong" into something you can partly see.

Read the public operational tracks, too. Not all scuttlebutt is a conversation. A plant running two shifts where it ran one, hiring pages that show a company staffing up (or a hiring freeze that contradicts a growth story), the volume of goods a business moves through public logistics and tax rails — these are observable facts that corroborate or contradict the narrative without anyone breaking a confidence. In India, aggregate signals of activity — the pace of dispatch, the buildup or run-down of stock in the trade — can often be inferred from public and observable sources, and read against management's claims they are a powerful cross-check. The rule is simple: it must be information you can lawfully observe or that is already public, never a confidential number pulled from inside.

Weigh by independence and pattern, not by volume or vividness. The most dangerous input in scuttlebutt is the single vivid anecdote — the one furious ex-employee, the one gushing dealer — because a vivid story feels like proof while being a sample of one. The method's output is only as good as the independence of its inputs: three dealers the company introduced you to are one managed message; three you found yourself, in different cities, are three checks. A pattern that holds across genuinely independent sources is a signal; a loud voice repeated, or a set of sources all steered to you by the company, is noise dressed as confirmation. You are looking for the mosaic — many small, individually-unremarkable public facts that together resolve into a picture — not for one big secret.

Everything above is lawful, and one thing is not, and the difference is the most important sentence in this module. Scuttlebutt is the assembly of public and observable information — the empty shelf, the two-shift plant, the dealer's own view of his own sales — into a mosaic that no insider handed you. The instant a piece of that picture becomes a specific, price-sensitive fact that the company has not yet made public — next quarter's unpublished results, an unannounced order win or loss, a pending acquisition — you have left research and entered material non-public information, and trading on it, or passing it to someone who does, is insider trading: a criminal offence under SEBI's regulations, not a grey area and not a clever edge.

Two practical rules keep you safely on the right side. First, judge the fact, not the feeling: ask whether the information is already public, and whether on its own it would move the price if it were. If it is non-public and price-moving, it is MNPI however innocently it reached you — and the correct response is not to use it, not to trade around it, and not to pass it on. Second, never solicit a breach: it is legitimate to ask an ex-employee about their own past experience; it is not legitimate to ask any insider for confidential numbers, and a source who volunteers a specific unpublished figure has handed you a liability, not an edge. The mosaic theory — that combining immaterial public pieces into a material conclusion is lawful — is the whole basis of scuttlebutt, and it collapses the moment any single piece is itself material and non-public.

The question scuttlebutt answers is constant — is the story the accounts and the management tell actually true on the ground? — but the door you knock on to answer it inverts by sector, because the signal lives in a different place in each business.

The inversion is that the same instruction sends you to a completely different place in each business, and the source that carries almost all the signal in one is absent or useless in another. Tell a reader to "check demand on the ground" for a two-wheeler maker and the whole weight falls on the dealer and the channel — off-take versus dispatch, stock ageing in the trade — because that is where genuine demand separates from channel stuffing. Carry that same instinct to a hospital and there is no dealer; the demand signal is patient flow, seen first by referring doctors and by the clinical staff who leave. Carry it to an IT-services firm and there is no physical channel at all — the truth lives in attrition and in what large clients are doing, which only ex-employees and customers can tell you. A reader who learns one sector's channel as the channel will run a diligent check against an irrelevant source and come away confidently misinformed — the store-check reflex applied to a software business, the dealer call applied to a road-builder. The method survives the move across sectors precisely because it tells you to find the source with no incentive to flatter and the clearest view of the real activity; which source that is, you must relearn for each business.

Take a composite mid-cap consumer-durables maker — call it Meridian Appliances — that reported a striking year. Revenue up 34%, the management commentary all about a hit new product line and "unprecedented channel enthusiasm." On the headline it is a fine result. But the forensic read from earlier in the guide flagged a gap: receivable days stretched from 58 to 96, and inventory at the company rose only modestly while revenue jumped — which means, if the product is really selling, the stock must be piling up somewhere downstream. That is your thesis, and it is a question, not a verdict: is this real off-take, or product pushed into the channel to book the sale? illustrative

So you go and look, from the accounts outward. You visit a dozen appliance retailers across three cities — chosen by you, not by Meridian's investor-relations team. Nine of the twelve are carrying far more Meridian stock than they want, several complain the company's salesmen pushed hard to load them before year-end, and two mention generous credit terms that let them take stock they cannot yet sell. You find, through a public professional network, a regional sales manager who left in the last year, and — asking only about his own past experience — he describes a culture of quarter-end "primary sales" targets met by shipping to distributors regardless of secondary off-take. None of these, alone, proves anything; together, and independent of one another, they resolve into a pattern that breaks the management's story rather than confirming it: the 34% is significantly dispatch into a channel that is now full, and the stretched receivable is the loan that financed the loading. illustrative

Now hold the discipline. The same exercise could have confirmed the thesis instead — if the shelves were empty, the retailers were reordering, and the ex-employee described genuine pull — and an honest reader would have accepted that answer just as readily, because the value of the method is that it can come back either way. Notice, too, what you did not do: you did not ring a friend in Meridian's finance team and ask what next quarter's number would be. That would have been the one move that turns this from research into a crime, and it would have been the least useful move as well, because a leaked figure tells you what the company already knows, whereas the empty-or-full shelf tells you something the company is trying not to say. The habit to build is to send scuttlebutt out with a specific thesis, to weight the answers by the independence of the sources, to accept a broken thesis as gladly as a confirmed one, and to stop dead at the legal line every single time.

Scuttlebutt confirms or breaks a thesis; it does not, by itself, produce a reliable measurement. A dozen store checks are a sample, not a census, and a sample you gathered is prone to every bias in your own head — you remember the vivid complaint, you notice the shelf that fits your fear, you stop calling once you have heard what you expected. The output is a weight of evidence for or against a thesis, read alongside the accounts, never a precise number that overrides them. Treating "I visited some shops and they were quiet" as if it were a measured decline in sales is small-sample bias wearing the authority of fieldwork, and it can be more misleading than the filing it was meant to check, precisely because it feels first-hand and therefore true.

Nor can scuttlebutt protect you from yourself if you run it to win. The deepest hazard is not a bad source but confirmation bias: the pull to seek, notice and remember only the evidence that agrees with the view you already hold. A reader who wants the growth story to be true will find the enthusiastic dealer, dismiss the sceptical one as an outlier, and call the result confirmation; a reader who is short the stock will do the mirror image. Honest scuttlebutt requires the one thing the method makes hardest — deliberately hunting for the evidence that would break your thesis, and giving a genuine independent dissent more weight than a chorus of agreement, because in a world where the company can steer you to friendly voices, the unsteered dissent is the rarer and more valuable signal. If your scuttlebutt has never once changed your mind, it is not research; it is a ritual.

And scuttlebutt cannot see what the sources themselves cannot see, or will not say. A dealer knows his own shelf, not the national picture; an ex-employee carries a grievance as often as an insight, and may be wrong, out of date, or settling a score; a competitor will tell you a rival's weakness and hide his own. Every source is partial and many are motivated, so the method is a triangulation of unreliable narrators, valuable only when their errors are independent and cancel rather than compound. It raises the cost of being fooled by a self-report, and it occasionally delivers the one fact that reframes everything; it does not deliver certainty, and a reader who mistakes a handful of confident anecdotes for the truth has simply swapped the company's bias for his own.

The first way people get fooled is by running scuttlebutt backwards — starting from a tip and going to find a company for it, instead of starting from the accounts and going to test a thesis. A whisper arrives — "there's a big order coming", "the promoter is buying" — and the reader assembles a case around it, calling the confirming pieces "channel checks." This is not research; it is a rationalisation with fieldwork attached, and it is doubly dangerous because the original whisper is often either worthless or, worse, material and non-public. The defence is direction: scuttlebutt is only legitimate when it flows from a reconciled question in the filings outward to independent evidence, never from a rumour inward to a justification.

The second way is mistaking agreement for confirmation without checking independence. Three dealers agreeing feels like proof, but if the company introduced you to all three, or if they all buy from the same super-stockist repeating the same line, you have one source counted thrice. The company that wants to manage its scuttlebutt will happily supply a roster of enthusiastic contacts, a curated plant visit, a reference customer — a whole managed picture that looks like independent confirmation and is a single message. The remedy is to weight by genuine independence: sources you found yourself, in different places, with different interests, and above all the dissenting source the company did not want you to reach, which is worth more than any number of steered agreements.

The third way is the vivid anecdote and its cousin, the small sample treated as data. One furious ex-employee, one dramatic store visit, one gushing distributor — each is emotionally loud and statistically silent, and the mind promotes the vivid to the general without noticing. A reader hears one scathing account of a company's culture and concludes the business is rotten; another sees one packed store and concludes demand is booming. Both have let a sample of one overrule a reconciled reading of the accounts. The discipline is to hold every anecdote as a single data point, to seek the pattern across independent sources rather than the loudest single voice, and to remember that the plural of anecdote is not data — a pattern is data, and a pattern is what you are actually looking for.