How to Read Management Guidance and What It Leaves Out

Run it as a procedure, not as a reading of tone. Go to the document the company filed rather than anybody's summary of it, separate every committed number from every adjective, write down the measures nobody addressed, date each statement and note the basis it was made on, then compare it at the result on the same measure and the same definition. One period settles nothing, so score each statement separately.

Every step below is an instruction about where to look and what to write down. None of them explains what guidance is, how the filed version differs from the spoken one, or how a forecast gets built from either. Guidance, the two versions and the building of a forecast are settled elsewhere and assumed here. The order below can be run on the morning a company reports, and it finishes with a sheet somebody else can check line by line.

The order is what keeps an impression out of the reading, so the order matters more than any single step. An analyst who listens first and sorts afterwards has already formed a view, and the sorting then arranges itself around the view instead of testing it. Sort first, on the words alone, and the view has to arrive later and argue with a list. The delay is the whole design.

Step one: where do the actual words come from?

From the company, in writing, and in three places. The primary documentThe statement a company filed itself, rather than anybody's report, note or article about it. A document that did not come from the issuer or the venue it filed with is not primary. is the results filing, which goes to the exchanges alongside the numbers. Beside it sits the released presentation, prepared for the same day. Then there is the transcriptThe written record of what was said on a results call, including the questions asked and the answers given. Companies commonly post one within a few days of the call. of the call, which is longer, less prepared and where most of the adjectives live.

All three are pulled. The three will not say the same thing, and the differences between them are information available no other way. A wedding caterer illustrates the point. There is the written quote, there is the sheet of options they left on the table, and there is the long phone call in which a cousin was told not to worry about the paneer. All three are real. Only one of them will be produced when the bill is disputed, and only the phone call contains the reassurance everybody actually remembers.

Step two exists to make a split. A summary has already made that split, and usually made it wrong. Somebody else decided which sentence was the headline, which adjective counted as guidance and which question was worth reporting. Every one of those was a judgement, none of them was recorded, and none can be undone from the summary. Starting from the filed words means starting from nobody's judgement but the company's.

Step two: how are the statements split?

Two columns, and nothing else at this step. On the left go the committed statementsA statement that names a measure, a period and a bound. A later result can then show it to be wrong. The difference between the two kinds of guidance is settled separately.: a measure, a period, a bound. On the right go the directional statementsA statement that names a measure and attaches an adjective to it, with no number or range. A directional statement points somewhere without ever being falsifiable.: a measure and an adjective, with nothing to fail against. Read each sentence, decide which column it belongs in, move on.

Whether a statement is plausible, ambitious or likely does not belong at this point. Judging plausibility is a different job for a different afternoon, and doing it here contaminates the sort. A tailor who says the blouse will be ready by Tuesday evening has said something he can be held to on Wednesday morning. A tailor who says it will be ready soon has not, however sincerely he meant it, and the sincerity is not the thing being sorted.

Before any opinion about the speaker enters the room, the split is done on the words. Run the year three communication of Sarvani Coatings Limited, an invented paints manufacturer, through it. The left column takes exactly one entry, high single digit volume growth. The right column takes one, gross margin holding around the prior year level. Two sentences are the entire harvest of a results day: one that can ever be shown wrong, and one that cannot.

Step three: how is what was not addressed written down?

The analyst takes the profit ladderThe ordered list of lines running from revenue down to profit after tax, with the costs subtracted in a fixed sequence. Its construction is covered in the accounting material. and goes down it line by line, ticking each one that was mentioned and writing down every one that was not. Then come the lines that sit outside the ladder but still carry the year: capital spend, the working capital movement, the segment split, the effective tax rate. The unticked lines are the silence.

Writing the silence down is the step people skip, and it is skipped for a structural reason rather than a lazy one. Every other step in this procedure is prompted by something already in front of the analyst. A document cannot say what it left out, so step three is prompted by nothing at all. The silence is produced from the ladder rather than from the document. The silence has to be written deliberately, or it never gets written.

The same thing happens outside finance. A household sits down to plan the month and talks for an hour about the school fee and the festival shopping, and nobody says the word electricity. The bill still arrives. Absence never announces itself, and nothing in the conversation flagged the missing bill. For Sarvani Coatings' year three, two of the eight lines that carry the year were addressed and six were not. A quarter of the ladder was spoken about, and three quarters left in the dark.

Step four: why does every statement need a date and a basis?

Because a statement is only comparable with a result that was measured the same way, and both the date and the basisHow a figure was put together: which items were grouped where, and under which policies. Change any of that and a different thing has been measured under the old name. are what make that check possible. Beside each statement go the day it was made, the period it covers, and the definitions in force at the time: which lines were grouped where, how the segments were drawn, whether a figure was reported or adjusted.

The reason is entirely practical. Definitions move. A company changes how it draws its segment reportingThe breakdown of revenue and profit by business or geography that a company publishes alongside its consolidated figures. How segments are drawn is a policy the company sets and can revise. boundaries, or restates a cost from one heading into another, and the line that carries a name in the statement is no longer the line carrying that name in the result. A statement made before a change in definitions cannot be compared with a result reported after one, and the date attached at this step is the only thing that makes the mismatch visible at all.

A shopkeeper quotes a price in March, and in June the same item costs more, but the quote covered delivery and the June price does not. Neither figure is dishonest. The two figures are answers to two different questions wearing the same label, and only the date on the quote signals that a check is needed.

Step five: how is a statement compared against the result?

Put the statement and the result side by side, on the same measure, the same definition and the same period, and write both numbers down without editorialising. For Sarvani Coatings' bounded statement the comparison runs: high single digit volume growth, taken at a floor of 7.0 per cent, against volume that came in at 6.0 per cent. One percentage point short. The growth rate was measured from the prior year revenue of Rs 2,120 crore. Valued on that base, the missing point is Rs 21.2 crore of revenue that did not arrive.

For the adjectival statement the comparison runs the other way. The adjective pointed at the prior year gross margin, 44.0 per cent, and said the measure would hold near it. Gross margin came in at 46.0 per cent, 2.0 points above. On year three revenue of Rs 2,415 crore, holding at 44.0 per cent would have given gross profit of Rs 1,062.6 crore. The published figure is Rs 1,111 crore. The difference is Rs 48.4 crore of gross profit above what the adjective pointed at.

The same shortfall measured against a different base is a different number, and no reader can tell which base was used, so the base a comparison was taken on is always stated. The one percentage point of volume is Rs 21.2 crore on the year two base of Rs 2,120 crore and about Rs 24.2 crore if taken on year three revenue instead. Both arithmetic is correct. Only one of them answers the question asked, and writing the base beside the figure is what lets somebody else find out which.

Step six: why is each statement scored separately?

Because the two statements were never about the same thing, so any figure combining them measures nothing that exists. Volume growth and gross margin sit on different lines, are driven by different forces and were made with different degrees of commitment. One column of the sheet takes the volume observation. The other takes the margin observation. There is no third column.

A single number is easier to carry into a meeting and easier to compare across companies, so the temptation to produce one score is strong. Resist it. Any school report makes the point: a child scores 30 in mathematics and 95 in drawing, and the average of 62.5 describes no child in the room. Both real facts vanish into it, and the one useful instruction, get help with mathematics, vanishes with them.

Netting a Rs 21.2 crore shortfall against a Rs 48.4 crore surprise produces Rs 27.2 crore and destroys the information in both. The Rs 21.2 crore is revenue that did not arrive on a statement management chose to put a bound around. The Rs 48.4 crore is gross profit that did arrive, above a level management was cautious about. The shortfall and the surprise are two findings pointing in opposite directions, and that is the most useful shape a set of observations can have. The netted figure has neither of them in it.

Step seven: what does a record across periods give?

The chance, eventually, to say something about the people speaking. Nothing before step seven permits it. A single sheet holds one row per statement per period, carrying the measure, the bound, the result, the base and the definitions in force. The sheet is added to every period. Beside the sheet, written in advance, sits what would change the reading of the speaker. The reading is decided before the evidence arrives, not after.

One year of Sarvani Coatings gives two observations: a bounded volume statement that came in below its floor, and an adjectival margin statement that came in two points above the level it pointed at. Two observations are not a record of anybody. Five years of the same two measures would give ten observations. If definitions moved in one of those years, the two rows for that year are voided rather than adjusted, leaving eight comparable ones. Everything in steps one to six produces facts about statements, and only a run of periods long enough to show a habit produces anything about the speaker.

A vegetable seller is not judged on one morning's tomatoes. The judgement rests on whether the crate at the back matches the crate at the front, week after week, through a good season and a bad one. A single visit tells only about that visit. Only the repetition, on the same stall and the same vegetables, becomes knowledge about the person.

What does the whole procedure look like, run end to end?

The complete run on Sarvani Coatings Limited's year three communication follows. Not a single row of it holds an opinion about a person.

Each entry is either a document a colleague can open for themselves or a division they can do again on the back of the same envelope. There is one test that matters for work of this kind, and this is it: could somebody handed the sheet rebuild every line without needing to ask a single question? There is nothing behind the word confident to rebuild, so a sheet containing it fails that test immediately.

Notice too what the run refuses to do. The run does not decide whether missing the volume floor was excusable, whether beating on margin was skill or a gift from input prices, or whether Sarvani Coatings' management is candid. All three are real questions. None of them is answered by anything above, and the run is more useful because it stops cleanly at the boundary rather than drifting across it.

One more thing about the run. Six of the seven steps are sorting operations performed on words, and a slider cannot sort a sentence. One genuinely numerical relationship remains: a bounded statement measured against the result it is scored on. Step five works it through with the base written beside it, and that division is the whole of the arithmetic this procedure ever needs.

Who actually runs this, and on what morning?

A buy side analyst runs it on results morning, before the price has settled and before anybody in the room has said the word confident out loud. The sorted sheet goes into the file with the date on it, and its real value arrives two years later, when somebody asks whether this management has a habit of guiding cautiously on the measure they later beat.

A credit analyst at a lender runs a narrower version, caring less about the bounded statements and much more about the silence. The lines that were never addressed are usually the lines a lender cares about: capital spend, the working capital movement, how much of the year's cash actually stayed in the business. For Sarvani Coatings, all three of those sat in the silence, and a lender who noticed that has a specific question to ask rather than a general unease.

A household investor with three shares and a Sunday afternoon runs the shortest version of all, and gets most of the benefit. Open the filing, find the two or three sentences that name a measure, write them on the back of an envelope with the date, and look at the envelope next year. The discipline costs nothing but the willingness to write things down before an opinion has formed. More resources make the procedure faster, never better.