How to Read an Annual Report: The Order That Makes It Work

An annual report is read in an order, and the order is the method. Seven steps: establish what is being held, read the auditor's opinion, read the accounting policies, read the statements, read the notes that matter to the question at hand, read the narrative last, and write down what is still not known. Reading it front to back means meeting management's framing before the numbers. The framing quietly decides what gets noticed.

Start with a scooter rather than a filing. Two people go to look at the same second-hand scooter on the same Sunday, and both spend twenty minutes on it. The first listens to the seller: one careful owner, serviced on time, only ever used for the school run, and then he glances at the papers and asks a mechanic to take a quick look. The second asks for the registration papers first, then the service book, then has the mechanic look at it, and only after all of that does he stand and listen to the seller. Same scooter, same seller, same twenty minutes of work. The two men walk away with different lists of things to check, and the difference is not diligence. The first man was handed his questions. The second man arrived with his own, and used the seller to answer them.

An annual reportThe document a company publishes once a year carrying its audited financial statements, the notes behind them, the auditor's report and a written account from its directors. In India a company files one whether or not it is listed. is that scooter at a larger scale, and the same twenty minutes becomes roughly an ordered hour. Everything in it is true, everything in it was checked to a standard, and the parts of it are bound in an order that was chosen for a printer rather than for a reader. The remedy is not a technique for reading harder. The remedy is a sequence, and changing the sequence changes the output even when the reading itself is identical.

A reader arrives here holding a great deal already. The three statements have been met more than once, along with what sits in the notes, what a written management commentary is doing, what a segment disclosure shows, what an ownership disclosure reveals, what an adjusted number can and cannot be trusted to say, and how a statutory report differs from the voluntary documents that grow up around it. Each of those is treated in its own right elsewhere. The order those sections go in is the one thing left, and each section is pointed at rather than opened again.

In what order is an annual report read?

Seven steps, and they are numbered because the numbers are the point. Each step has an output, and the output of one step is what makes the next step readable. Work them in this order even when the document is bound in a different one. The document always is.

The order is not a preference about where to start, it is what decides which questions a reader finishes holding. The right-hand column carries the point. Every step hands the next step something, and by step seven what is held is not a conclusion at all. Step seven hands over a list. The list is the intended output, and a reader who expected a verdict will think the method failed at exactly the moment it worked.

Why does the order matter more than the reading?

Here is the uncomfortable part. The reading work is identical on both paths. The same sheets get turned, the same figures get looked at, the same amount of attention gets spent. Nothing about the front-to-back reader is lazier or less careful. The one difference is which part of the document was met first, and that difference alone is enough to change what the hour produces.

The mechanism is ordinary and human. Whatever is read first supplies the categories used to sort everything after it. Meet a written account of the year before the numbers, and the numbers arrive already labelled: this one is the growth, that one is the investment, this other one is the seasonal effect. Meet the numbers first and they arrive unlabelled. Unlabelled is uncomfortable, and it is exactly the state in which a reader generates their own questions. Reading front to back is not a slightly inefficient version of the same activity, it is a different activity with a predictable output.

The difference can be counted. Walking the same report twice on Anjani Stationers Private Limited, once in the procedure's order and once front to back, the two readings finish with lists that overlap in only four places. The procedure's order finishes with thirteen questions, nine of which the front-to-back reading never asks. The front-to-back reading finishes with nine, five of which are the narrative's own themes handed back to it in the form of a question. Neither list is dishonest. Only one of them was written by the reader.

What is established before anything else is read?

Step one takes about four minutes. Because it feels like admin rather than analysis, step one is also the one people are most tempted to skip. Four things, each written at the top of the note sheet. Which entity is this. Are the figures standalone or consolidated. Which period do they cover and which period are they being compared with. And has any comparative figure been restated since it was last published.

On Anjani Stationers, step one returns this. The entity is Anjani Stationers Private Limited, a maker of school notebooks and exercise books. Both a standalone and a consolidated set are present, the standalone carrying assets of Rs 1,80,00,000 against equity of Rs 1,42,00,000, the consolidated carrying assets of Rs 2,09,50,000 against equity of Rs 1,59,50,000. The period is the second year, compared with the first. No comparative was restated, so no figure in the prior-year column has moved since it was published. Four minutes, four facts, and now every later number has an address.

Step one comes first in every reading procedure worth having because everything downstream means something different when step one is wrong. Reading the consolidated equity of Rs 1,59,50,000 while thinking the standalone is in hand puts the reading Rs 17,50,000 out before a single real question has been asked. Miss a restatementA correction of a previously published figure, so that the comparative column in this year's report no longer matches what was printed last year. It has to be disclosed, and it changes what a year-on-year comparison is measuring. and every growth rate computed afterwards is measuring the correction as well as the trading. Neither of those is a subtle error. Both are invisible unless somebody looked.

When is the auditor's opinion read, and why then?

Step two is short, it takes about three minutes, and it is the step readers most often get wrong by omission rather than by mistake. Readers do read the opinion. Because that is where the opinion fell in the binding, they read it last, or somewhere in the middle. A reader who reaches the opinion last has read the whole document without knowing whether any of it could be relied on.

Two things make the opinion worth its position. The opinion is the only part of an annual report written by somebody outside the company, and it is the part that says whether the rest is fit to be read at all. Everything else in the document, every statement, every note, every line of commentary, was written by the business about itself. Single authorship is not a criticism, it is simply what a report is. The auditor's opinionA short statement from an independent auditor saying whether the financial statements give a true and fair view. It carries its own defined forms of wording, and it may be accompanied by paragraphs drawing attention to particular matters. is the one place where somebody else takes a view, and reading it first is what settles how much weight the following hour can carry.

Step two returns two things and no more. Whether the opinion is in its ordinary form or something other than ordinary, and whether anything sits beside it drawing attention to a particular matter. On Anjani Stationers the report carries an opinion covering the statements that follow. The position of step two is what carries the lesson, not the contents of this particular opinion.

Why do the accounting policies come before the numbers they define?

Step three takes about ten minutes and it is the step most readers skip entirely. Step three is also the step that most often changes a conclusion. The reason is mechanical rather than clever. A policy read afterwards cannot un-read a number. Once a closing stock figure has been looked at and an impression formed of it, learning twenty minutes later which cost formula produced it does not remove the impression, it only adds an awkward footnote to it.

The accounting policiesThe stated choices a business has made about how it measures things: how stock is costed, how assets are depreciated, over how many years, and how a lease is treated. They sit near the front of the notes and they apply to every figure printed after them. are not background material. The policies are the definitions of the numbers. On Anjani Stationers, step three returns four sentences and no more: the cost formula for stock is first-in-first-out, depreciation runs on a straight line, the four classes of asset carry lives of eight years, eight, four and three, and a term is assumed for the warehouse lease. Four sentences make a ten minute read, and after it, the inventory of Rs 28,00,000 and the property, plant and equipment of Rs 36,00,000 mean something specific rather than something general.

What happens to the statements at step four?

Step four is the longest single stretch, about twenty minutes, and it has one discipline attached to it that is easy to state and hard to keep. All three statements get read, in whatever order the question at hand suits. By step four the order among the three no longer matters. And as they are read, every place where two things that ought to move together did not gets written down, and then the writing stops.

Stopping the writing is the discipline. A divergenceA place where two figures that would ordinarily move together did not, such as a stock balance growing much faster than the sales it supports. It is a question to be raised, never a finding on its own. gets listed at step four and explained at step five or six, never at step four. The temptation to explain on the spot is enormous. A plausible explanation always presents itself, and the moment one is accepted the search stops for the note that would have tested it. Step four's output is a list, not an explanation, and writing the explanation here is the commonest way to lose a reading that had been done properly up to that point.

Anjani Stationers gives four divergences at step four, and all four are already familiar. Receivables grew 21.8 per cent while revenue grew 12.5 per cent, from Rs 2,40,00,000 to Rs 2,70,00,000. Inventory grew 47.4 per cent. The gross margin held at 45.0 per cent in both years while the operating margin fell by 6.7 points. And the provision against receivables tripled, from Rs 3,00,000 to Rs 9,00,000. Alongside those, the cash flow statement shows operating cash of Rs 36,30,000, investing of minus Rs 34,00,000 and financing of minus Rs 4,30,000. Together those three took cash from Rs 7,00,000 down to Rs 5,00,000. Five things noticed, nothing concluded, and the list goes into step five.

Which notes does step five open?

Step five takes about fifteen minutes and it has two halves. The first half is driven by the reader's own list: for every line that matters to the question brought to the document, the note behind it gets opened. The second half is fixed and does not depend on that question at all. Three notes get opened every single time, whatever the reader came to find out: related parties, commitments, and contingencies.

Why those three and not others. Because they are the only notes that routinely carry amounts no line on the face of the balance sheet points at, so a reader who navigates purely by the face will never arrive at them. On Anjani Stationers they hold Rs 21,20,000 between them: a warehouse commitment of Rs 10,80,000 over three years, a guarantee of Rs 8,00,000 over the borrowing of Chitra Binding Works, and a disputed claim of Rs 2,40,000 from the Sunrise Public School group. Separately the related party note shows Rs 8,00,000 of binding invoiced by Chitra Binding Works during the year, of which Rs 1,50,000 was still unpaid at the year end. Three notes read regardless of the question are what stop a reading from being shaped entirely by the question the reader happened to arrive with.

Why is the narrative read last?

Step six takes about eight minutes, and it is the whole reason the other six steps are in the order they are in. By now the reader holds a list of their own questions, and turns to the narrativeThe written account of the year that management publishes alongside the statements. In a listed company it usually appears as a formal management discussion; in a private company it sits inside the directors' report. with that list in hand, and it gets checked against the list, question by question. Which of those questions does this answer. Which does it answer in a way that can be tested against something already read. Which does it walk past without mentioning.

Read in that position, the narrative is enormously useful and completely safe. The narrative is management saying what it thinks happened, at a moment when the numbers are already known, and the gap between the two accounts is information available no other way. Read first, the same words do something entirely different: they hand over the categories, and every number met afterwards gets sorted into one of them. Read last, the narrative is an answer to the reader's list. Read first, it is the list.

On Anjani Stationers the narrative sits inside the directors' report. A private company does not publish a management discussion in the listed sense. The company produces no investor presentation and holds no earnings call, so there is no third document to go to and nothing to reconcile the report against. Because there is exactly one written account of the year and it is the last thing read, the position of step six is easier to see rather than harder. The directors' report is read against the four divergences, with one plain question: whether it addresses them.

What gets written down at step seven, and how is the stopping point known?

Step seven takes five minutes and produces the only thing kept. Step seven writes down what is still not known, and beside each item the evidence that would settle it. Not what is suspected. Not what it probably is. The open question, and the document or the figure that would close it.

The stopping rule is the seven conditions listed beside the steps, and it is a checklist rather than a feeling. Stop when the basis is established, the opinion has been read, the policies are understood, the divergences are listed, the relevant notes have been opened, the narrative has been checked against the reader's own list, and the residueThe set of questions still open after a reading is finished, each one written down with the evidence that would answer it. It is the output kept, and it is expected rather than a sign of incomplete work. is written down. Seven conditions, all satisfied, and the reading stops. Not because curiosity has run out, but because the document has stopped being able to answer.

Now the part that unsettles people. A residue of open questions is the expected output of a complete reading, not evidence that the reading was incomplete. An annual report is a summary prepared by a business about itself, checked to a standard, and published to a format. The document was never built to close every question a careful reader can raise, and a genuinely good annual report leaves four or five of them standing. A reading that produces no residue at all has usually done one of two things: stopped before step five, or answered its own questions with the narrative's explanations and written them down as findings.

What must never be a step?

Three things get done routinely, and none of the three is a shorter version of this procedure. Each is a different method that produces a different output, and that is why all three are named rather than treated as slips.

The first is reading the narrative first. Reading the narrative first is the whole subject of step six and the reason the sequence exists, and its output is a question list written by the business rather than by the reader. The second is treating another document as a substitute for the report: a slide deck, a results call, a summary somebody else prepared. A deck, a call and a summary are all real and useful, and they are also selected. The report is the one document that has to carry what the format requires whether or not the year went well. Anjani Stationers, being private, produces no presentation and holds no call. The substitute is not a lesser version of the report, it is sometimes not available at all.

The third is concluding from a single divergence. Receivables grew 21.8 per cent against revenue up 12.5 per cent, and there is a conclusion sitting right there waiting to be picked up. Do not pick it up. A single divergence is one input to a question list and never an output on its own. Certainty is not something an annual report is built to produce, so a reader who finishes one certain about a business has almost certainly stopped too early.

What does the whole procedure look like run on one business?

An ordered hour, start to finish, on Anjani Stationers Private Limited. Every figure below has already been published and not one of them moves here. Watch the time column as much as the output column. The shape of the hour is part of the method: four minutes of admin, three minutes of somebody else's opinion, ten minutes of definitions, and only then twenty minutes on the numbers.

Sixty five minutes, and look at what is not in the last row. There is no verdict about whether this is a good business. There is no view about whether the provision was set correctly. There is no judgement about anybody's conduct, and there could not be. Nothing in the sequence was designed to produce one. The whole reading takes an ordered hour and produces a question list rather than a verdict, and the question list is the thing carried into the next conversation.

The settings read out as follows. Walking the procedure's order to step seven closes the list at thirteen questions, with all seven conditions of the stopping rule satisfied. Walking front to back to step seven closes the list at nine, with the basis never established and the policies never read, so two of the seven conditions can never be met however long the document is sat with. At any step, the strip in the middle counts the same three things: what only the procedure's order has asked by now, what both paths have asked, and what only the front-to-back walk has asked. The shared column never rises above four, and every question in the front-to-back walk's first step is a theme lifted straight from the narrative it opened with.

Who reads an annual report in this order, and what do they do with it?

Three people open the same document in the same week and none of them is doing what the others are doing, but all three run the same seven steps.

A lender runs the seven steps because the order decides how much the security is worth. Step one tells the lender whether the assets being looked at are the standalone Rs 1,80,00,000 or the consolidated Rs 2,09,50,000. The distinction is not a detail when the money is being advanced to one of the two. Step five is where the lending question actually gets answered. The Rs 8,00,000 guarantee over the borrowing of Chitra Binding Works and the Rs 10,80,000 warehouse commitment are obligations that sit outside the Rs 38,00,000 of recognised liabilities. A lender who read the balance sheet and stopped has priced against Rs 38,00,000 when Rs 21,20,000 more was disclosed in the notes behind it.

An analyst runs it because the output is the deliverable. The five open questions at step seven are the email the analyst sends, and it is a far better email than a view would be: specific, answerable, and impossible to argue with. Nobody can dispute a request for next year's ageing table. Plenty of people can dispute an opinion about a provision, and the analyst who sends the opinion instead of the request has spent their credibility on something the document never supported.

And Vaidehi Rao, as finance controller, runs it in reverse, on her own report, before it goes out. She walks the seven steps as an outsider would and writes down the five questions the sequence produces. Then she checks whether the narrative in the directors' report answers them, and where it does not, she writes the sentence that does. Running the reader's own order against a company's own report before publishing it is how a narrative stops being a set of themes and becomes an answer to the questions the numbers actually raise. The reverse reading takes an hour once a year, and it is the cheapest hour in the whole reporting cycle.

Anjani Stationers Private Limited, Chitra Binding Works, the Sunrise Public School group and Vaidehi Rao are invented.
Educational material. Not advice on any investment, tax, budget or market position.