Guidance: What Is Committed and What Is Merely Hinted

Guidance is what a listed company says in advance about measures it chose to name, and it is a communication rather than a measurement. The formal version the issuer files is one thing. The directional version somebody speaks on a call is another. The analyst's work is separating a committed number from an adjective, and writing down everything that was never guided at all.

Almost every mistake made with guidance comes from one small slip of grammar. People say the company guided to nine per cent as though nine per cent were a fact about next year. It is not. Nine per cent is a sentence somebody said. Sentences can be careful, cautious, hedged, self serving or simply wrong, and none of that shows up once the sentence has been boiled down to a number and pasted into a spreadsheet. So the first move is to put the sentence back.

What is guidance, and who is actually speaking when it is given?

Guidance is a statement about a future period, made in advance, by or on behalf of a listed company. The definition is complete there, and notice what is missing from it. Nothing about accuracy. Nothing about obligation. Nothing about how much of the business it covers. A guidance statement is a communication, and like every communication it was composed by somebody who decided what to put in it.

Three separate choices sit inside every guidance statement, and all three are readable before a single number is compared: the measure, the period and the precision. Somebody decided which line of the accounts to speak about. Somebody decided whether to speak about a quarter, a year or the medium term. And somebody decided whether to attach a bound to it or leave it as a description. The measure, the period and the precision are information in themselves, and all three are available on the day the statement is made, long before any result exists to check it against.

Statements are already read this way in ordinary life. A painter quoting for a flat says the job will take about eight days. A different painter says it will be finished by the fourteenth, and if it is not, the last two days are free. Both of them have said something. Only one of them has said a thing that can turn out to be false in a way anybody could point at. And neither of them said anything at all about what happens if the plaster underneath is worse than it looked. The plaster is the part that actually decides whether the job runs over.

Company Guidance: the version the issuer files and has to stand behind

Company guidance is the formal version. The issuer itself makes the statement, in a filed document or in a released investor presentationThe slide set a listed company releases alongside a result, summarising the period and often carrying its own forward statements. Where and how it is filed is a matter for the exchanges., and it is attributable to the company as an institution rather than to any individual. The filed statement has been through whatever internal process the company runs before something goes out with its name on it. The original can be found, its exact wording read, and its release date seen.

The filed version is the only one the company itself has to answer for later. Its wording is usually far more careful than anything written about it afterwards. Careful is not the same as informative. A filed statement is often narrower and vaguer than the summary that follows, precisely because somebody sat with it and removed everything that could come back. Where the filed original and the report of it are read side by side, the difference between the two is frequently the most interesting thing on the desk that morning.

The regulator sets what a listed issuer must disclose about a future period, whether it must say anything at all, and what it has to do if a statement stops being true. In India the regulator is the Securities and Exchange Board of India (SEBI). Where a filed statement and a released presentation actually sit is a matter for the exchanges, and both are named at the end.

Management Guidance: the version somebody says out loud

Management guidance is what individuals say. Ravindra Setlur, the chief financial officer of Sarvani Coatings Limited, an invented paint maker, takes a question on an earnings callThe meeting a listed company holds after publishing a result, where management presents and then takes questions, usually with a transcript released afterwards. How to read one is taken up under earnings calls. and answers it in his own words, at speed, without a drafting committee. He might be more forthcoming than the filed document. He might be less. He is certainly less precise. Ordinary speech is less precise, and a question he did not expect is being answered in about eleven seconds.

Almost everything quoted as guidance is this second kind, and almost every summary of it treats the two as one thing. That conflation is the single most common error in this whole area, and it is easy to catch. The question is where the sentence came from. If the answer is a filed document, it is the first kind. If the answer is a call, an interview, a conference or an analyst meeting, it is the second, and the record should name whose mouth it came out of rather than attributing it to the company.

Neither kind is superior. The spoken version is often where the useful colour lives. A person answering a question will explain reasoning that no filed document would ever carry. The point is not to prefer one. The point is to know which one is in hand. The two kinds carry different weight, they are found in different places, and only one of them is a statement the company has committed itself to.

How is a committed number told apart from a hint?

The test runs on the words, not on the tone. Confidence in delivery tells nothing at all. A committed statement carries three things: a measure, a period, and a bound. A hint carries an adjective. The measure, the period and the bound are the whole test, and the check takes about four seconds.

Here are the two statements Sarvani Coatings made at the start of year three. First: management expected high single digit volume growthThe change in the number of units sold, with every effect of price and mix stripped out of it. How volume and price are separated inside a revenue line is worked through under volume and price.. Second: management expected gross marginGross profit stated as a percentage of revenue, so revenue less the cost of materials, divided by revenue. Gross margin is the first rung of the profit ladder and was settled in the accounting material. to hold around the prior year level. Read them side by side and the difference jumps out. The first has a measure, a period and a bound. The second has a measure, no stated period, and an adjective where the bound should be.

A bounded statement can be shown wrong by a single published number and an adjective cannot, and that asymmetry is the entire difference between the two. Volume grew 6.0 per cent, which is at or just below the bottom of high single digit, so the first statement was missed and anybody can see it. Gross margin went from 44.0 per cent to 46.0 per cent. Did that hold around the prior year level? Two points is a large move on a gross margin. Because the sentence never said what around meant, it still cannot be called false. The second statement was not dishonest. The second statement was never capable of being scored.

Why does what was not guided matter more than what was?

Because the guided list is short and the unguided list is long, and the year is decided by the long one. Two measures were spoken about. Look at what was not. Revenue itself was never guided. Earnings before interest, tax, depreciation and amortisation, or EBITDAOne rung further down the profit ladder than gross profit. What sits inside the measure and what does not was settled in the accounting material., was never guided, and it rose 31.2 per cent. Capital spend of Rs 186 crore was never guided. The Rs 54 crore that the working capital cycle absorbed was never guided. The split between decorative and industrial was never guided. The effective tax rateTax charged for the year divided by profit before tax, which is often different from the headline statutory rate for reasons the accounting material sets out. of 25.1 per cent was never guided.

The unguided list is where the year is usually decided, so a reader who writes the silence down has already learned more than a reader who scores the guide. Most treatments leave the unguided list out entirely, and leaving it out turns guidance from evidence into a substitute for thinking. A guided measure has been chosen for the reader. An unguided measure has been chosen against, and the reason it was left out is a question worth asking every single time.

A builder's quote for a room works the same way. The quote says the tiling will be finished in nine days. The quote says nothing about what happens if the wiring behind the wall turns out to be old, nothing about who pays for the extra skip, and nothing about the two days lost when nobody can get the fittings. The customer is given one bounded promise and half a dozen silences, and everyone who has ever had work done knows which of the two decides how the job actually goes.

How is a guide checked against the result afterwards?

Four things have to match before a comparison is a comparison. The same measure: volume against volume, never volume against revenue. The same definition: reported against reported, adjusted against adjusted. The same period: the year that was guided, not the twelve months that happen to read better. And the same basisThe definitions and accounting policies a figure was prepared under. If a policy changed between the guide and the result, the two numbers are measuring slightly different things even when they carry the same name.: the accounting the guide was prepared under, unchanged.

The check fails on definition far more often than on arithmetic, and when a definition has moved the honest response is to say so and stop, rather than adjusting until the numbers fit. Suppose a company guided on adjusted EBITDA and now reports reported EBITDA. A bridge can be built between the two. Presenting the bridged figure as a score of the guide supplies an adjustment nobody guided and quietly buries the mismatch inside the adjustment. The words the definition moved are the ones to write, naming what moved, and the guide is left unscored for that year. An unscored guide is a finding. A scored guide built on a mismatch is a fabrication with a decimal point on it.

What was each of the two statements actually worth?

The next step is to stop treating them as sentences and put rupees on them. Almost nobody takes that step, and it is the one that changes how guidance is read for good. The instinct comes first, before the arithmetic.

Start with the published year. Revenue was Rs 2,415 crore against Rs 2,120 crore the year before, a rise of Rs 295 crore or 13.9 per cent. The rise came from two things: volume up 6.0 per cent and realisationRevenue divided by units sold, so an average selling price across everything a company shipped. Realisation moves with both price changes and shifts in what was sold, and the two are separated under volume and price. up about 7.5 per cent. Multiply the two factors together, 1.06 times 1.07467, and the answer is 1.1392, the 13.9 per cent. The year rebuilds exactly.

The margin statement sizes the same way. Gross profit at the delivered 46.0 per cent on Rs 2,415 crore of revenue is Rs 1,110.9 crore. The company published Rs 1,111 crore after rounding. The prior year level the guide pointed at was 44.0 per cent. The 44.0 per cent had produced Rs 933 crore of gross profit on Rs 2,120 crore of revenue. Had margin genuinely held there, gross profit on the larger year three revenue would have been Rs 1,062.6 crore. The difference is Rs 48.3 crore, or two points of margin on Rs 2,415 crore of revenue. Comparing the published Rs 1,111 crore instead of the unrounded Rs 1,110.9 crore gives Rs 48.4 crore, and the tenth of a crore between those two figures is rounding rather than anything real.

Then size the volume statement the same way. Take the bottom of high single digit as 8.0 per cent. Hold realisation exactly where it landed and rebuild the year: Rs 2,120 crore times 1.08 times 1.07467 gives Rs 2,460.6 crore of revenue. The rebuilt revenue is Rs 45.6 crore more than the company delivered, and at the 46.0 per cent margin it actually earned, those lost sales were worth Rs 21.0 crore of gross profit.

The statement made most precisely turned out to be worth less than half the statement management was cautious about: Rs 21.0 crore against Rs 48.3 crore, a ratio of about 2.3 to one. The reason is not mysterious once it is seen. On this revenue base one point of gross margin is worth Rs 24.15 crore of gross profit. One point of volume is worth only Rs 10.48 crore. Both statements moved by exactly two points. The two statements were never worth the same, and no amount of care in the wording of the volume guide was going to change the gap.

With the control pushed down towards 3.0 per cent, something worth seeing happens. The two effects cross over at a volume outcome of exactly 3.5 per cent, and below that the volume shortfall is the bigger of the two. The crossover point does not depend on the size of the company at all: 46.0 per cent applied to 1.08 equals 48.0 per cent applied to 1.035, so the revenue base cancels straight out of the comparison. The lesson is not that margin always beats volume. The lesson is that neither statement can be ranked until both have been sized, and precision of wording is no guide whatsoever to size.

What does a missed guide and a beaten one in the same year reveal?

Very little, and saying so plainly is harder than inventing a story that fits. There is one year. In that year, one statement was missed and one was beaten. There are at least three stories that fit those facts equally well, and the published record separates none of them.

A single period cannot separate caution from information, so anybody who reads one outcome as evidence about the speaker has drawn a conclusion from one observation. This is not scepticism for its own sake. The objection is plain arithmetic about evidence. One period gives one data point on each statement, and a data point is not a pattern. The reader who writes management is conservative on margin after one cautious remark that turned out well has performed the same move as somebody who calls a batsman a good player of spin after a single innings.

The alternative is boringly simple and almost nobody does it. The statement is recorded, in the words it was said. It is dated. At the result the four match tests are run and one line written down: missed, met or beaten, and by how much. Then it is done again next period, and the period after. After three or four years there is a record of how this particular management speaks about the future, and a record like that is worth having. Until then there are observations, and they should be called observations.

What can guidance never reveal, whatever it says?

Three things, and every one of them gets assumed constantly. Guidance does not reveal what management actually knows. A statement is what they chose to say, not what they believe. Guidance does not bind them either. Circumstances change, and a company that revises a guide has done something ordinary rather than something shameful. And its absence is not evidence about the business.

A company that gives no guidance at all has said something about its disclosure policy and nothing whatsoever about its year. This one is worth dwelling on because the misreading is so tempting. Silence feels like it must mean something. Surely a confident management would say so. In fact companies decline to guide for reasons that have nothing to do with the year ahead: a policy that predates the current management, a view that guiding invites short term pressure, a genuinely unforecastable input cost, or simply a preference for saying less. A policy of silence may fairly be thought unhelpful. Silence cannot be read as a signal about the coming twelve months.

How does an analyst use this on an ordinary Tuesday morning?

Meghna Iyer covers coatings and Sarvani Coatings is on her list. When the results release lands she does not start with the numbers. She opens a two column note. On the left she writes what was guided, in the exact words, with the date and the place it sat. On the right she writes everything the year turned on that nobody spoke about in advance. The right hand column is longer every single time, and she has never once found a year where it was not.

Her rule is that no line enters a model attributed to management unless she can point at the sentence and the document it came from, and everything else carries her own name. The rule sounds bureaucratic. The rule takes four minutes. When a forecast turns out wrong, she can find the assumption that broke without re-reading a year of work. She already knows which lines were hers and which were somebody else's.

A lender does a narrower version of the same thing. Presented with a borrower's projection, the useful question is never whether the projection looks reasonable. The useful question is which parts of it the borrower has actually committed to, in a document, with consequences, and which parts are the borrower being optimistic in a meeting. A household does it too, without any of the vocabulary: when the shop next door says the delivery will come Thursday, the household plans around Thursday, and when it says things should ease up next month, it does not. Everyone already separates commitments from adjectives. The only difference here is that the sentences are longer and the rupees are bigger.