Business Analysis: Understanding a Company Before Valuing It

The financial outcome sits downstream of the business, so analyse the business first. Three ordinary operating decisions at one invented register maker, about people, about space and about a binding operation bought into, raised the standing bill Rs 24,40,000/- and carried the operating margin from 22.08 per cent to 15.37 per cent. Not one was a decision about a margin, and the split beneath that figure is an estimate rather than a disclosure.

Revenue rose and the margin fell nearly seven points. So where did the profit go?

Begin with a record rather than with a definition. Anjani Stationers Private Limited, an invented register maker, is where the record starts. Paper goes in at one end, hard-bound registers come out at the other, and the registers go to schools. Two of its years are published in these notes, and between those two years something happened that most readers name wrongly on sight.

Revenue went from Rs 2,40,00,000/- to Rs 2,70,00,000/-, so it rose. Operating profit went the other way, from Rs 53,00,000/- to Rs 41,50,000/-. Each of those set over its own revenue gives an operating marginOperating profit set against the revenue of the same year, written as a percentage. A margin is a division and nothing more.. In the first year 53,00,000 over 2,40,00,000 gives 22.08 per cent. In the second, 41,50,000 over 2,70,00,000 gives 15.37 per cent. The margin fell 6.71 points in a year when the business sold more, not less.

Now shut the door everybody reaches for first. If a trade turns nasty on price, the damage lands in the contribution marginThe share of each rupee of sales still standing after the costs that sale itself caused, and before any of the bills that arrive whether or not anything is sold. before it lands anywhere else. A contribution margin takes off only the costs each sale drags in with it and nothing else. At Anjani Stationers that figure read 42.75 per cent in the first year and 42.78 per cent in the second. The contribution margin did not move. Three hundredths of a point in either direction is not a price war; it is arithmetic noise.

So the reader is left holding two published facts that will not sit comfortably together. More registers went out of the door at the same effective price, and less profit came back. Something between those two sentences swallowed Rs 11,50,000/-, and no percentage anywhere in the accounts is going to say what.

The shape is familiar from ordinary life. A household gets a pay rise in April and finds the month tighter in November. The tightness was a new rent agreement signed in July and a set of school fees that started in June. The payslip records the rise perfectly and will never explain either of them. A payslip is not defective for that. A payslip is simply not the document where causes live.

What decides whether a figure handed over by somebody is worth anything?

Before working out where the Rs 11,50,000/- went, one thing has to be settled: what decides whether any answer reached is worth writing down. A fact carries only the authority of the route it came down, so research is the route and not the number. The claim does more work than it looks like, so earn it rather than accept it.

Take two research sheets carrying the same question about a trade. On the first, a line reads not established. On the second, the same line carries a trade average, with nothing beside it saying where the average came from. Today, the second sheet is the better looking document. The second sheet is complete and reads as finished work. In a year it is the worse one, and the difference is not a matter of taste.

The line reading not established stops being blank the day anybody lays hands on the fact. The blank line also tells whoever reads it next exactly what to go and ask for, so it does two jobs at once. In a year nobody will remember that the average was supplied rather than found, so the filled line can never be corrected. The filled line has stopped being questionable. The second line already has a name worth borrowing whole: an invention wearing the clothes of research.

Everyday version, and it is exact. Two people write down what the flat upstairs pays in rent. One goes up the stairs and knocks, and comes back with a number and the name of the person who said it. The other takes the going rate for the building off a listing site and writes it in. Both sheets now carry a rent figure. Only one of them can be argued with, and only one of them will still be right after the tenant upstairs renegotiates.

So what should go on the line instead? Three short commands settle it, in the original words.

Writing the blank down is not a counsel of despair; it is the only version of the line with a future in it. It costs the appearance of completeness and buys the ability to be corrected, and on a sheet somebody will still be quoting in a year, that is a good trade.

How to Analyse a Business Before Analysing Its Financials: what runs first?

Here is the order of work, and it is an order rather than a checklist. The difference matters, so hold on to it: a checklist asks whether every box is filled, and an order of work asks whether each step was allowed to come back empty. A seven-step run over exactly this trade is already published, and the frame behind running a field in a fixed order belongs to Michael E. Porter, who is named for the ordering and for no figure whatsoever.

The run is worth copying for a reason other than its list of headings. Every step is written out with four things attached: the evidence it goes looking for, the condition under which it stands, the condition under which it does not, and the instruction it hands downward when it does not. A failure is never a dead end in that design. A failure is a message to the step underneath, and so the run keeps moving even after three lines in a row come back holding nothing.

The opening step records two things: what trade this actually is, and which side of it the analyst has taken up position on. Leave either one out and the failure is total. The step underneath has nothing to work against and simply stops. From there each step in the middle sets down, in advance and in writing, what would let it stand and what would sink it, and each sinking hands the work on rather than leaving a hole.

The last step is where the discipline bites. The final step gathers the five readings into one place and counts them. Its condition for standing is exact, and here it is in full.

Run over this field, that procedure returned a result most readers would file as a failure, and these notes published it as the answer. Two of the five readings stand. Three lines are recorded not established. One named buyer, the Sunrise Public School group, holds the terms, so the buying side stands. The terms sit at one small counterparty, so the input side stands. Rivalry does not stand. Nothing published says how Anjani Stationers and Bhavani Register Works behave towards each other, and nothing states how many makers the city holds. A newcomer does not stand either. Nobody has published what building or buying a works of this kind costs, and no maker is recorded as having arrived or left. Alternatives do not stand either. Nothing published names anything beyond registers that a school would put the same money towards.

Notice what did not happen there. Nothing was added, averaged or boiled down into one word. The count itself is the verdict, and it is written as a count: two of five, and three of five. A step permitted to return nothing at all is doing the work of a test. A step that always returns an answer is not. These notes put the same thing more sharply.

The counter-example is familiar from any counter. A form that will not be accepted until every box is filled establishes exactly one thing: every box is filled. The form establishes nothing at all about whether anybody knew what to put in the boxes. A procedure whose every line always fills in has quietly stopped being a procedure and become a form, and a form gets filled in.

The three empty lines are therefore worth more than they look. Empty lines like these are not an unfinished draft but next year's shopping list, written in the order somebody would go and get them.

Why does the order run that way round, and not the other?

The order of work has now been asserted twice and earned nought times, so earn it here, on the two published years. Three questions, asked in a fixed sequence, and the sequence is the whole lesson.

Question one, did the price side move? The contribution margin read 42.75 per cent and then 42.78 per cent. So no, it did not. The price question is settled from the accounts alone, cleanly, in one line.

Question two, did the bottom line move? The operating margin read 22.08 per cent and then 15.37 per cent. So yes, and by a great deal. The bottom-line question is settled from the accounts alone too, just as cleanly.

Sit with the fact that those two answers do not sit comfortably together. The discomfort is the finding rather than a defect in the reading. Nothing squeezed each sale, and yet a great deal less survived the year.

Question three, where did it go? Into the standing costA cost that does not wait for a sale. Space taken on a lease, somebody on a monthly salary, wear on a machine already standing in the works. base. The standing bill rose Rs 24,40,000/- between the two years, on three things in particular: people, space, and a binding operation the business bought into. And here is the part worth noticing. The accounts answered question two perfectly and could not answer question three at all without a line that sits in a note and a line that sits nowhere.

The point generalises well beyond registers. A division gives the size of a movement. A decision gives its cause. The two live in different documents, and only one of those documents is audited. A shopkeeper's till roll records every sale of the year to the rupee and will never record that the road outside was dug up in March. The till roll is not wrong. A till roll was never the place where March lived.

How Business Economics Affect Financial Outcomes: what did three decisions do?

One instruction about that Rs 24,40,000/- is short enough to reproduce whole.

Naming where the money sits is the second half of the work, and everything below unpacks it.

The rise sits in three places. Pay went up by Rs 6,00,000/-. Whatever part of other operating costs stands still whether or not a register sells went up by Rs 11,40,000/-, most of that being a second warehouse the business took across the year. Wear and write-down on what it had bought went up by Rs 7,00,000/-. Put the three together and 6,00,000 plus 11,40,000 plus 7,00,000 lands on Rs 24,40,000/- precisely, leaving nothing over to allocate anywhere. And the label goes in the same breath as the figures, here and everywhere below that names them: that three-way division was somebody working it out, not the business stating it.

The finer reading is better than the coarse one, so take it now. Employee benefits expense is a prescribed headA line a set of published accounts is required to present under its own name, rather than one the business chose to show or group as it pleased., printed on the face of the statementStanding as its own line on the statement, instead of being tucked into the notes behind it or left for the reader to put together.. So is depreciation and amortisation. Two of the three parts are therefore nothing more than movements in those two named lines, and 6,00,000 plus 7,00,000 comes to Rs 13,00,000/-. Over 24,40,000 that is 53.28 per cent of the rise. Both can be read off, and read off again by a second person tomorrow.

The middle part cannot be read off anything at all. Dividing a cost by how it behaves, into what moves with each sale and what stands still, is simply not a thing any statement is asked to show. Nobody withheld it and nobody was ever required to publish it. So the finding is sharper than the version most readers land on: the widest slice of that total, the 46.72 per cent, was never published by anyone, and the total coming out perfectly is exactly what keeps that fact out of sight.

Then the claim this whole section exists to land. Not one of those three decisions was a decision about a financial outcome. Taking on people is a decision about work getting done. Taking a second warehouse is a decision about where stock sits. Buying into a binding operation is a decision about a stage of production. The treatment where the division was first published says so directly.

The margin did not fall because anybody chose a margin. The margin fell because three people in three meetings solved three different problems, and the accounts added their answers up afterwards without being asked.

Households do this constantly and never call it economics. In one year a household takes on a second scooter for the commute, moves to a flat one floor larger, and starts paying somebody to mind the children in the afternoon. Three decisions, three separate reasons, not one of them about the monthly surplus. Between them they account for the whole of what happened to the monthly surplus. Nobody wrote the split down at the time, so nobody in the household could say which of the three did how much.

One thing is easy to say and hard to feel: that a rise in what stands still comes straight off what survives. Everybody agrees with that sentence and then, five minutes later, reads a fallen margin as a fact about the trade. So before the panel below moves anything, predict what it will show.

The arithmetic closes with nothing left over. So what has it proved?

The reconciliation is worked in the open. How persuasive it is should register before anything is said about what it is worth. ContributionWhat one sale leaves behind after the costs that sale itself caused. It is the money available to meet the bills that turn up regardless. rose from Rs 1,02,60,000/- to Rs 1,15,50,000/-, a rise of Rs 12,90,000/-. The standing bill rose Rs 24,40,000/-. Take one from the other and 12,90,000 less 24,40,000 is minus Rs 11,50,000/-. Now look at what actually happened to operating profit: Rs 53,00,000/- down to Rs 41,50,000/-, a fall of Rs 11,50,000/-. The same figure, to the rupee, with nothing left over for any other cause.

The closed loop is a genuinely satisfying object, and it is also the exact moment the awkward thing has to be said. The figure that closed the loop rests on a split that was built rather than read, so a constructed split cannot confirm the level it was built out of. Both sides of that subtraction were assembled from the same set of accounts using the same estimated division of cost by behaviour. The two sides could not have disagreed. Agreement between two things that share an input is not a check; it is arithmetic being consistent with itself.

So there is a test, it takes a minute, and almost nobody runs it. Set the ingredients of each route side by side, in full, and see whether one quantity turns up on both lists. If it does, the two routes were never separate. The word carrying the damage is always independent, and it goes unexamined because two numbers agreeing feels like the checking has already been done.

A published example makes the trap concrete. Three separate builds of one marketplace each came to rest on the very total they had been divided out of in the beginning, and not one of the three checked anything. A headcount multiplied by that same total over a headcount can only ever give the total back, so each loop shut on the figure it had opened from.

Two people check a shopping bill. Both of them add up the same printed list of prices and both get the same answer. Agreement establishes that neither of them made a mistake in addition. Agreement does not establish that the shop charged the right price for anything on the list, and no amount of re-adding will ever reach it.

Now the honest close. The modest claim is still a real one. The arithmetic does locate which term the movement came out of, and that is worth having: it puts the movement in what stands still rather than in what moves with each sale, and it rules out the trade story that most readers would have written. Locating the term is a weaker claim than the closing of the loop makes it feel, and stating plainly which of the two claims is being made is the whole of the discipline.

Which column did each of these three figures come out of?

The discipline set out at the start now applies to the three decisions, one at a time. Beside each of the three stands its route.

People. Employee benefits expense is a prescribed head sitting on the face of the statement, so the movement of Rs 6,00,000/- is read, and the route is the statement itself. Who was taken on and to do what is not there and never was.

Space. The second warehouse taken during the year is named in prose where the split was published, so its existence is read. How much of the rise it accounts for sits inside the Rs 11,40,000/- component, and that component was estimated rather than disclosed. So one half of this line is read and the other half is built.

The binding operation brought inside. The stage is named. The stage carries no separate price in the published figures. The movement in depreciation and amortisation of Rs 7,00,000/- is read off a prescribed head, and the connection between that movement and the binding work is inferred rather than stated anywhere.

A reader who cannot say which column a figure came from cannot say what would change it. The bite is practical. If a figure was read, it changes when the business publishes a different one. If it was built, it changes when somebody revisits the assumption, and so it can change without anything at all happening in the business. Read and built are two entirely different kinds of instability, and they need telling apart before anybody leans on the number.

And where a line cannot be filled from a route that can be named out loud, it reads not established, and that line is exactly what next year's work goes and gets.

The same number under two headings. So how does one word destroy it?

Close on the object that shows the whole argument in one move. Nobody makes an arithmetic error in it and nobody invents a number, and the figure is destroyed anyway.

Against a rated capacity of 4,00,000 registers in a year, the works turned out 2,50,000, so it ran at 62.50 per cent of that rated capacityThe yearly output a works is designed for, assuming its machines hold their stated rate through the hours available. A fact about the plant and about nothing else.. Put the idle part through at the same realised priceThe average amount per unit that actually reached the seller once every discount and allowance came off, worked out by dividing money received by units sold. of Rs 108.00/- and revenue would come to Rs 4,32,00,000/-. Filling the works and holding the price puts the operating margin at 25.65 per cent against the published year at 15.37 per cent. Moving from 2,50,000 registers to 4,00,000 is a rise in volume of 60.00 per cent, and it is a rise in volume and nothing else at all.

Every one of those figures is correct and every one is published. Now put a heading on the slide. Under SUPPLY CEILING AT THE SELLER'S OWN PRICE, the figure holds, so long as two things go with it. One, that sorting cost by behaviour was a judgement somebody made and not a thing the accounts stated. Two, that the price is pinned at Rs 108.00/- right across a year 60.00 per cent bigger in volume than the one published. State those two and the number is honest, a fact about machines and a price.

Now change the heading to ADDRESSABLE MARKET and change nothing else at all. The same digits, in the same order, on the same slide. Except that the new heading asks for three things nobody has: how many buyers there are who want a register of any kind, whether one of them would pay this particular price for this particular register, and where the edge of that group is drawn. A works has no opinion at all about whether the world wants what it makes.

A jug of water measured correctly to the millilitre is a correct figure. Labelled with how thirsty the street is, it is destroyed, without the water being touched.

Here is the part worth sitting with, and it inverts what most people expect. Being published and correct is what made the mistake hard to spot, not easy. A slide whose arithmetic reconciles to the operating accounts wears the look of something already checked, and in one sense it is. A slide with an obviously invented number invites a challenge from the first reader who looks at it; this one does not.

What does a first-hand document leave behind, and what does a first-hand act leave?

Three separations, one clause each, and then the work is handed on.

The order in which one document outranks another as evidence, and what makes a record an original one in the first place, are both settled at length elsewhere in these notes, and used here rather than rebuilt.

Management's choices, and how a reader tells a choice apart from a consequence, form a different question from where the evidence for either came from. Somebody deciding to take a second warehouse and somebody working out afterwards what the warehouse did to the margin are two separate acts, and only the second one belongs here.

And the separation this whole reading order exists for, stated once and worked through in full under the subjects named below. Something being first-hand is a property of a document, and doing something first-hand is a property of an act, and the two cross rather than nest. A published statement is an original record that a second reader can open on their own, and reaching it means working from a record somebody else made. An answer given by the person who actually knows is as first-hand as anything gets, and it leaves behind no document for anybody to open, so it has to carry a name and a date instead of a citation.

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