How to Write a Business and Industry Case Study

A case study wants a story with an ending, and published evidence supplies exposures rather than events. So the honest output is a record instead of a narrative: every line carries where it came from, a statement, a walk outside, or a labelled construction, and a line with no route is struck out. The record closes with what is known and what is not established, both written the same size.

A year of this business is written up and every figure adds up. So where is the ending supposed to come from?

The difficulty only shows itself once there is something concrete in front of the reader, so a record comes before a procedure. Anjani Stationers Private Limited, an invented manufacturer, turns paper into hard-bound school registers and sells them to schools across one district. Two of its years are published, and the movement between them is small enough to hold in mind through the rest of this guide.

ContributionWhat is left out of revenue once every cost that grows and shrinks with output has been paid for. went from Rs 1,02,60,000/- to Rs 1,15,50,000/-, a rise of Rs 12,90,000/-. The standing baseSpending that arrives in the same size whichever way volume goes: the shed, the people on salary, the wear on machines the business already has. went from Rs 49,60,000/- to Rs 74,00,000/-, a rise of Rs 24,40,000/-. Operating profitWhat the year's trading left behind before interest and tax were taken out of it. went from Rs 53,00,000/- to Rs 41,50,000/-, a fall of Rs 11,50,000/-.

Set those three movements one under another and do the subtraction yourself. The subtraction runs to one line and it settles something. Rs 12,90,000/- of extra contribution against Rs 24,40,000/- of extra standing cost leaves Rs 11,50,000/- unaccounted for, and Rs 11,50,000/- is exactly what the profit line lost. The movement closes with no residueThe part of a movement that the named causes fail to account for. It is nought when the arithmetic closes., so there is no room inside it for a fifth cause that nobody named. Each year also stands up on its own. Take the standing base of Rs 49,60,000/- off contribution of Rs 1,02,60,000/- and Rs 53,00,000/- is what remains. Do the same in the later year, Rs 74,00,000/- off Rs 1,15,50,000/-, and Rs 41,50,000/- is what remains.

The sources named at the foot also record what the larger standing base went on. The extra spending went on people, on space, and on a binding operation the business bought into. Where each of those sits in the accounts is set out in those sources. The split behind the standing base is described there as an estimate rather than a disclosure, and that qualification returns later and never travels separately from the figures it belongs to.

Consider where a writer actually stops. A year of a business has been read, every figure reconciles in both directions, and perhaps three paragraphs remain before the end of the write-up. The question forms on its own, and it is the reader's question as much as the writer's. What happened next?

Answer it flatly. This evidence holds exposures and never events, so nothing published anywhere answers that question. The shape of the silence becomes clear once the absences are named one at a time. Nothing published records that the largest account left. Nothing records that a counterparty stopped supplying. Nothing records that a machine stood still, that an account was lost, or that a school changed maker. Every one of those would make a fine last paragraph, and not one of them exists.

The word that matters there is exposureSomething a business is open to, recorded with no statement at all about whether it will occur.. A business with thirty per cent of its revenue in one account is open to something. Whether anything ever arrives is a different sentence, and the second sentence is the one nobody wrote down. The missing second sentence is a property of the evidence rather than a gap in the writing, and a procedure that pretends otherwise produces one invented sentence per study.

There is an everyday version of this and it is worth keeping hold of. A household can produce every bill it settled last year, to the rupee, filed by month. Asked what the rent will be after the current agreement runs out, the same file has nothing whatever to say, not because anybody is holding it back but because that was never a transaction of theirs to record. Every rupee that passed through is written down completely. Every rupee that did not pass through is not written down at all, and no amount of rereading turns the second into the first.

Why does the shape of a document decide what ends up being invented in it?

Four things sit underneath everything that follows. Each one explains a step that would otherwise invite argument, so all four are worth setting out before any procedure starts.

The first. A fact borrows its standing from the route it came down and from nothing else, and every other rule in this guide follows out of that one sentence. Two lines can sit one under the other in the same document, in the same weight, in the same face, saying things of roughly the same size, and one of them can be tested by anybody with a telephone while the other cannot be tested by anybody at all. Nothing about how a line reads will show which of the two is in hand. Where it came from is the only thing that does, and a document that records that for no line is a document in which every line quietly borrows the standing of the strongest line it contains.

The second. A narrative cannot be corrected and a record can, and that is an engineering property rather than a matter of taste. In a story the claim is spread across the telling, so a reader who doubts it has nowhere to point; the doubt attaches to the whole thing or to nothing. In a record each line answers for itself, so the next person can fill one in, move one or take one out without disturbing anything around it. The test of a case study is not whether it reads well. The test is whether one line in it can be corrected by somebody who was not in the room when it was written.

The third follows from the second and it is the uncomfortable one. The shape of the document decides what gets invented, and it decides it before any research begins. A shape with a resolution slot in it will get a resolution, because the slot is drawn into the document and an empty slot looks like unfinished work. The slot is why what follows is a procedure whose output is a record rather than advice about writing carefully. Careful writing inside the wrong shape still fills the slot, and the slot is where nearly every invented sentence in this subject comes from.

The fourth is about position, and it is the reason the damage lands where it does. The last line is the one that survives. Nobody quotes the reconciliation in paragraph four. Everybody quotes the closing sentence. The closing sentence is what the document was for. A study in which forty lines carry their route and the final line carries none has put its only unsourced claim in the one position from which it will be repeated by people who never saw the working. The weakest line in a case study is usually the most travelled one.

What is a case study actually for, if it is not a story?

Most readers arrive expecting a story with a lesson at the end, and there is nothing foolish about that expectation. A story with a lesson is what the words case study have meant in classrooms everywhere. The trouble is not that a story is a lesser instrument. The trouble is that a story is a different instrument, built to do a job this evidence cannot support.

Make the difference physical rather than abstract. Hand somebody a story about a business that ran into difficulty and ask them which sentence to go and check. Nobody can answer, and the failure is not slowness: the claim lives in the arrangement rather than in any one sentence, so there is nothing to point at. Hand the same person a record and every line answers the question for itself. One line came out of a set of accounts. Another came out of a telephone call. A third was built to show a property and says so where it stands.

A case study is a document somebody will pick up a year later, and the only question that matters then is whether they can tell which lines are still true. That is a plain test and it is unforgiving. A year on, the reader has forgotten the argument, forgotten who wrote it, and forgotten what was in the news the week it was written. The lines are all a reader still has, and the lines that survive contact with a year are the ones carrying enough about themselves for a stranger to re-test.

Then say what the record buys that the story cannot. Somebody who has run a works for twenty years can take it apart a line at a time, and that is the only sort of disagreement that improves a document instead of merely ending it. A record also leaves the next reader holding a short list of errands, so finished work becomes the opening half of somebody else's.

The everyday version sits on any ordinary street. A neighbour's account of why the shop on the corner closed is memorable, complete and satisfying, and not a word of it can be checked. The shop's own order book is dull, partial and full of gaps, and every line in it can be tested against somebody else's copy of the same transaction. One of the two is worth repeating and the other is worth relying on, and the whole of this guide is about not confusing them.

What are the three routes a line can come down?

The three routes are worth taking one at a time, each with a test that can be applied in the moment rather than a definition that has to be remembered.

A statement is a document a business publishes about itself, recording transactions it actually made. A statement arrives complete and costs a download. A statement can also be exhausted. Exhaustion is the property people find hardest to believe until they have felt it: a reader can genuinely reach the end of a set of accounts, read every line, and know that nothing further is coming from that direction, however long anybody sits with it.

A walk is a fact obtained by asking somebody who does not work at that business. Somebody at a mill reading a rate down the telephone. A price list pushed across a counter by whoever was standing behind it. A head teacher answering a question about which maker she has heard of. A walk costs a telephone call rather than a download, and it arrives one answer at a time and on somebody else's schedule. There is always one more person who could be asked, so a walk cannot be exhausted.

A construction is arithmetic built to demonstrate a property rather than to describe a business, and it is the only one of the three that the writer makes rather than collects. The permission to build one is narrow and the way it goes wrong is specific, so a construction has rules of its own, set out further down.

Neither of the first two is more respectable than the other, and the third is not a lesser version of either. They answer different questions and they cost different things, and a line's standing comes from its route rather than from how impressive the route sounds. A shopkeeper asked what he paid for a sack of rice answers down to the paisa. The purchase was his own and the slip is in the drawer. Asked what the same sack fetches in the lane behind him, he has to put his slippers on and go and read the board. One answer is worth exactly as much as the other. The two answers are simply got in different ways, and pretending the second came out of the drawer is the whole of the fault.

Which brings the rule the rest of the procedure hangs on, and it is short enough to keep.

Every line carries where it came from, and a line with no route is struck out.

Struck out, not softened, and this is the instruction writers resist hardest, so it is worth spending a paragraph on. Softening a line changes its wording and leaves its status alone. The softened line stays in the document, it stays in the file, and next year somebody quotes it without the hedge. The hedge was the part that read like padding. A struck line is gone, and nobody can quote a line that is not there. The difference between the two is not tone; it is whether the sentence still exists to travel.

Why a document about one business cannot answer a question about another business's transactions is worked out in full and is covered separately under Company Analysis vs Industry Analysis: Where the Evidence Comes From. The procedure below takes that finding as settled and works from it.

What does the stamp look like once it is worked on a real business?

Descriptions of a discipline are easy to agree with and hard to apply, so what follows builds the record rather than describing it, marking every line as it lands. Everything below is published in one of the sources named at the foot of this guide. Nothing in it was worked out for the occasion.

Four lines came out of documents the business publishes about itself. The standing base rose from Rs 49,60,000/- to Rs 74,00,000/- while contribution rose from Rs 1,02,60,000/- to Rs 1,15,50,000/-, so operating profit fell from Rs 53,00,000/- to Rs 41,50,000/-. The book of buyers runs to thirty six accounts, every one invoiced directly. The largest is the Sunrise Public School group at Rs 81,00,000/- against revenue of Rs 2,70,00,000/-, or 30.00 per cent. The same group is Rs 38,00,000/- of a gross bookThe whole of what buyers still owe, counted before anything at all is set against it. of Rs 95,00,000/- still owed, or 40.00 per cent, a gap of 10.00 points against its share of the money earned. And the collection periodHow long the money due took to arrive, averaged and counted in days, worked back out of the accounts after the year closed. came to 171.23 days on that group and 110.08 days on everybody else, against invoices asking for sixty to ninety.

Three lines came from asking people outside the business. Somebody rang round the paper trade and found nine further mills sitting within delivering distance, every one of them making paper of the identical weight and finish, and every one able to put a price in front of a buyer inside a day. The same round of calls settled the second line. A mill never bought from before wants roughly a fortnight before its first load reaches the yard. The third came from walking into schools: across the district, head teachers recognise one of the two makers by name and do not recognise the other.

No set of accounts anywhere carries any of those three lines, and all three of them are facts. Both halves of that sentence have to stay together, because letting go of either one is where the trouble starts. Letting go of the first means an afternoon spent hunting for the mills inside a document that could never have held them. Letting go of the second quietly decides that whatever sits outside the accounts is soft, and throws away most of what the reader opened the write-up for.

One more line belongs in the record, and it goes in on purpose to show where a walk runs out. There is a second maker in the same line of work, building to the same specification: Bhavani Register Works, run by the people who set it up, turning out 1,50,000 registers a year. Nobody looked that up in anything. The count is the sort of thing anyone selling into a district picks up about whoever else sells into it. A walk like the rest, it carries its route like the rest. And in the same breath: that figure gives the size of one business and no size whatever for the field around it. Nothing anywhere states how many makers the city holds, so there is nothing for one maker's output to be a share of.

What goes on the line when the fact exists nowhere?

Put the three hardest questions to this business's own surroundings and count the answers honestly rather than hopefully.

How many makers does the city hold? Nothing published anywhere puts a number on that. One named second maker establishes that there are at least two, and at least two is a floor and not a count. And what does Bhavani Register Works charge for a register, and what does the paper cost it? Not one of those figures appears anywhere: not its price, not what it pays for paper, not what its works costs to run. And what would a school buy if it stopped buying hard-bound registers altogether? Nothing outside the trade is named anywhere at all.

Three questions asked and three refusals returned. And it is at precisely this moment that the temptation shows up, wearing the clothes of research. The phrase is the trade averageA number people in a trade repeat as normal for it. It counts as a source only when the person or the body that put it in print can be named., and one test decides it. Can the person who put that number in print be named? Where the answer is yes, it is a source like any other and the line stands. Where the answer is no, the phrase is doing a source's work without being one, and a line finished that way is indistinguishable in the document from a line finished properly.

The instruction that comes out of all this runs to three lines and does not need decorating. The question goes into the write-up. Beside it goes the one thing that would answer it. Then comes a plain statement that nobody has that thing.

A line reading not established can be completed on the day somebody finally gets hold of the fact. A line completed with an average nobody published can never be corrected at all, and a year from now nobody will remember that it was made up. That lopsidedness is the whole argument and it repays being said twice in different words. The honest line has a future in it: it is an errand for the next person, and the list of them gets shorter every year as people work through it. The invented line has no future at all. Such a line is finished the day it is written, it is never tested by anybody, and repetition alone hardens it into something everyone assumes was always known.

Now land this same fault where it is sharper. Here the average genuinely exists and is genuinely published. The other thirty five accounts came to Rs 1,89,00,000/- between them, or 70.00 per cent of the year, averaging Rs 5,40,000/- each, and that is 2.00 per cent each on average. The word average is not decoration there. The split inside the average is published nowhere at all, so there is no second-largest account in this record, no grouping of the top five, and no ranking of the book. Reading the average as any one account's share manufactures thirty five figures in a single stroke, and every one of them would look as solid as the Rs 81,00,000/- above it.

Say the plain thing about that too. The temper of the work depends on it. Nobody concealed the split. The split was never published. Nearly every book of buyers anywhere stands the same way, and that is the ordinary condition of the evidence rather than a black mark against anybody. A record that reads every gap as somebody's fault will be wrong most of the time and disagreeable all of it.

One last point about how these lines are drawn, and it is a formatting decision that turns out to be an argument. The not-established lines are drawn at the same size as the filled ones. A blank set in smaller type at the end of a write-up reads as an afterthought, something the writer ran out of time for. The same blank set at full size, in the body, with its two companion lines beside it, reads as a finding. A finding is exactly what it is.

When may the procedure build its own arithmetic, and what must it say in the same breath?

Sometimes the honest answer to the step before is that a demonstration is needed and no published figure supplies one. A reader who has just been told to strike every unsourced line will want to know whether that leaves any room to work anything out, and it does, but the room is narrow and its shape is worth learning exactly.

Here is the permission with its conditions, and take it in one breath rather than as a list to be skimmed. The arithmetic is labelled as a demonstration of a property inside the very sentence it appears in, and it names no country, no year, no line of work and no enterprise, and it is fastened to none of the businesses the write-up is about, and the published figures standing beside it keep their own labels so a reader can tell in one glance which is which.

Now the addition this subject needs, and it is the important half of the step. A construction may not supply a likelihood. The bar sits on the quantity itself rather than on where the figure came from, which is why labelling does not rescue it. A demonstration that says how probable something is has supplied the one input that decided its own answer. The label makes that worse rather than better. A labelled invention reads as an invention somebody has already dealt with.

A construction may supply a setting instead. A movement in revenue of a stated size. An account of a stated size no longer buying. A stoppage of a stated number of weeks. Say what would follow at that setting, work it through so a reader can check the working, and then say nothing whatever about whether the setting arrives. The setting is permitted and the chance of the setting is not. The sources this guide draws on already work exactly that way and say so in their own words: a distance is measured, and no claim is attached about whether anybody will travel it.

The everyday version settles it in one sentence. Measuring whether a table will fit through a doorway is a fact about the doorway and the table, and a tape settles it in a minute. The tape says nothing at all about whether anybody intends to carry a table through it this year, and a builder who wrote the second thing on the same sheet as the first would have mixed a measurement with a guess and given them the same ink.

What is the one shape of story this evidence genuinely supports?

Having refused the ending, it would be mean to leave the reader with nothing, and there is no need to. One real before and after is available, it is fully published, and it is a better ending than the one the shape was asking for. Give it in full. A refusal that arrives with an alternative reads as discipline, and a refusal that arrives alone reads as poverty.

The movement from the earlier year into the later one is a genuine narrative by any test that could be applied to a story. Something changed. The change is named. The consequence is measured. And the arithmetic closes with nothing left over, a thing a story almost never manages.

The movement runs in the direction a write-up would actually set it down. The business committed Rs 24,40,000/- more before anything at all was sold. Contribution rose Rs 12,90,000/-, and that rise is real. A write-up that skipped it would be describing only half of what happened. Operating profit still fell Rs 11,50,000/-, and the two movements account for that fall exactly, with no third cause needed and no room for one.

Now bring in the reading that everybody wants and carry its qualification in the same breath, every single time, without one exception. The degree of operating leverage moved from 1.94 to 2.78. Underneath both readings sits a split of the costs into the part that moves and the part that stands still, and wherever those two figures appear that split is called an estimate rather than a disclosure. The same place adds that the Rs 74,00,000/- underneath the later reading will not stay at that level once a one-time cost of buying into the binding operation falls away. Neither remark is a garnish. Detach either one and the write-up printing the figure has quietly promoted an estimate into a disclosure.

Then the part that turns an observation into an ending. Hold the standing base at the earlier year's Rs 49,60,000/-, let revenue grow on its own, and the reading goes down from 1.94 to 1.75. Revenue growth by itself was pushing the reading the other way. So the Rs 24,40,000/- of extra standing cost carried the whole distance from 1.75 up to 2.78, and it carried the part that revenue growth had already given back as well. The result is a finding with a direction and a cause and a size. Every figure in it rests on the same estimate, and that estimate travels with all three readings or with none of them.

The only ending this evidence supplies is arithmetic that closes. Every line of it can be checked by somebody who disagrees with it, and the ending the shape wanted has no such line anywhere in it. Hand that record to a person who has run a works for twenty years and they can argue with the split, argue with what the base will do next, argue with whether the extra spending was worth doing at all. Nobody can argue with an invented last paragraph. There is nothing in it to take hold of.

Once every line carries a route, what order do the lines go in?

The procedure has run as far as it goes, and it has left a question standing in plain view. Every line now carries where it came from. The blanks are written full size. And the writer still has to decide what goes first.

Say why that is not a formatting decision. The line at the top is the one that gets read, the line at the bottom is the one that gets skipped, and the writer chooses which is which before anybody else in the world sees the document. Nothing in the procedure above chose it. Nothing in the evidence chooses it either. The evidence is a set of lines, and a set has no order until somebody gives it one.

One warning belongs here and the rest belongs elsewhere. The smallest number in a record is often the largest dependency, so an order chosen by size buries exactly the line that mattered, right at the foot, where the reader who opens the file at ten to six will never reach it. Burying the line that mattered is not a rare accident. Sorting by size does it every time, by construction.

A record's refusals matter in the same breath, and the next study is built on them. A record orders and never rates. Nothing in it carries a score, a grade, a colour or a word standing in for how serious a line is. A grade is the product of two quantities and this evidence publishes one of them, so a graded record has multiplied a measurement by an invention and printed the result in a column where it looks like the other measurements.

The work stops there, and where the reader can see it stopping. Which order, on which published column, and what happens to every other position once the column changes, is set out under The Business Risk Register: Recording What Could Go Wrong. Producing the lines and putting them in order are separate jobs, and settling the ordering in passing would decide a question that deserves a study to itself.

Where does a careful study go wrong, if every figure in it is right?

Which six lines travel with every case study, whoever is reading it?