Organisation Design and Transformation: What Structure Costs

Structure decides where a decision gets settled and how long it takes to get there, so it changes what a business can do before it changes any number. From outside almost none of it is visible: these notes publish two organisational facts across five businesses and no third. The cost of changing a structure turns up as an estimate, and the estimate says so.

A reader wants to know how a business is organised. What can actually be found out?

The honest version of the question is not a technical one, so it is best asked the way it would be asked out loud. Who is in charge of what here? How many people does anything have to pass through? Who has to say yes before the price on an order can be moved? Every one of those is a fair question, every one of them decides what it is like to deal with the business, and the answer to none of them appears in a set of published accounts.

The thing can be felt first without any accounts at all. A customer stands at a counter in a shop asking for a small discount. Nothing on the shelves says whether the person behind the counter can settle that, and nothing about the size of the shop says it either. The answer is found out by asking. If the answer comes back in four seconds, something real about how that business is arranged has just been learned. If the person picks up a phone and the customer waits, something equally real has been learned, and it was learned from the wait rather than from any drawing.

Now bring that to the businesses these notes teach with. Five of them appear across these notes, all invented, and every one of them exists to make some idea concrete. So before defining a single term, do the inventory. How much organisational description is actually published about any of them?

Fact one. Bhavani Register Works is promoter runRun by the people who started the business and who still hold most of its shares, rather than by managers appointed on behalf of outside holders., and that description comes with its own gloss of what the words mean: run by the people who started the business and who still hold most of its shares, rather than by managers appointed on behalf of outside holders.

Fact two. Anjani Stationers Private Limited, a maker of hard-bound school registers invented for these notes, holds 70 per cent of Chitra Binding Works Private Limited and appoints the majority of its boardThe group that takes a company's formal decisions and to which its managers answer.. The same disclosure is unusually careful in the same passage: it states that nothing was merged, that two boards still sit, that two bank accounts still run and that two sets of statements are still prepared.

The two facts are the whole of it. There is no drawing of any arrangement anywhere in these notes, for any of the five. Nobody is counted. Nothing says how many steps a decision passes through, nothing says who answers to whom, and nothing says which questions get settled where. Two sentences, five businesses, no third fact.

Before that reads as a complaint, name what makes it useful. A reader standing outside a business is in this position almost always, and it is the ordinary situation rather than a special failing. The documents a business publishes are built to report what it sold, what it spent and what it holds. The documents were never built to report where decisions get settled, so a reader who goes looking for that in them is not being obstructed. The reader is looking at the wrong document, and no version of the right document exists.

So what is a structure for, if it is not a drawing?

A drawing is what most readers expect: a shape with somebody at the top, lines coming down, names in the boxes. With that expectation set aside, the question is what such a drawing would actually say. A drawing would give the arrangement somebody meant to set up. The drawing would not give the place where the discount question asked at the counter gets settled, and that was the question.

A structure is a fact about where a decision gets settled and who has to be asked before it is, rather than a shape. Everything else in this guide falls out of that one sentence. For any decision a business makes over and over, such as whether to hold a price for one customer or whether to run one order ahead of another, two things can be asked. Where does that decision actually get settled? Who has to be asked first? The pair of answers is the structure, for that decision. The pair needs no boxes, and the pair can be stated for one decision at a time by somebody who knows nothing whatever about the shape of the rest of the business.

A procedure survives where a definition does not, so the definition becomes a procedure. Four answers, and always in this order. One decision, named specifically. The place it is settled. Who has to be asked before it is settled. How long that takes, from asking to answer. Run four times over four different decisions, that procedure yields more about a business than any drawing would have given, and every one of the four answers can be checked against somebody who works there.

There is a further reason the drawing feels like the answer and is not. An arrangement is generally set up because of something the business decided to do, rather than the business deciding what to do because of an arrangement, a direction of dependence Alfred D. Chandler Jr set out at length. Read that way, an arrangement is downstream of a commitment. An arrangement is a consequence, and a consequence recorded on paper drifts from the commitment that produced it.

A structure is answered one decision at a time, and a business sits at a different place on the axis for different decisions rather than at one place for all of them. The same shop may settle a small discount at the counter in four seconds and send a request for thirty days of extra credit somewhere else entirely, and both of those are true of the same shop on the same afternoon. Anybody who says a business is arranged one way, full stop, has flattened a set of separate answers into one that describes none of them.

How the two ends of that axis are told apart by somebody standing outside, and why holding most of another company's shares is a different fact from taking its decisions centrally, is covered separately under Centralised vs Decentralised Organisation. No figure carries across to it.

Decentralised Organisation and its opposite: where does the same question get settled?

Two words do most of the work in this subject, and both of them name a place on the axis rather than a kind of business. Take one decision and hold it still while the place it gets settled moves. Nothing else makes either word mean anything precise.

A centralised organisation, for a given decision, is one in which that decision is settled at one point for the whole business. Ask the same question in two branches on the same morning and the same answer comes back. The question travelled to the same place both times. A decentralised organisation, for the same decision, is one in which it is settled where the work happens. The answer comes back in seconds, and the two branches may answer it differently. Two different people answered it.

Notice what has just been described. A trade is not a preference. Settling at one point buys consistency and pays for it in time; settling where the work happens buys time and pays for it in consistency; and a business that wants both for the same decision is asking the axis for something it does not offer. The trade is closer to arithmetic than to opinion. The question either travels or it does not, and whichever it does costs the thing the other one buys.

The same trade is lived at a counter. Ten shops sit in one shopping centre. In the first, the person behind the counter can settle a small discount there and then, and the customer is out of the shop in a minute with an answer the shop next door might not have given. In the second, the same request goes to a telephone, the customer waits, and what comes back is exactly what would have come back in any other outlet of that business anywhere. No arrangement was ever seen. The customer experienced only a wait and a degree of consistency. A structure feels like nothing else from outside.

Neither end is better, and ranking them would turn a description into advice. A business selling one standard product to thousands of buyers may want the same answer everywhere and be content to wait for it. A business fixing a different problem in every order may need the answer now and be content that two orders got handled differently. The axis describes a trade; it does not contain a recommendation, and no reading of it produces one.

Now the honest part, before anybody asks for the case. Nothing published anywhere in these notes says where either business sits on this axis for any decision at all. The two definitions are worked properly above, and no case follows them. A case would have to be made up, and a made-up case is worse than none. The four answers are not in a document anybody filed, so somebody who wants to see the definitions land on a business will have to go and ask that business the four questions.

What do the two published facts actually settle, and what do they leave open?

A fact quoted and left sitting there is decoration. Work each of the two. Each settles something narrower than it first sounds and leaves open something wider.

Take the promoter-run fact first. The promoter-run fact settles who the people taking the decisions answer to, and the answer is themselves and each other rather than holders outside the business. Knowing that is real and useful, and it changes how a conversation with that business goes. Then notice what it does not settle. The same fact says nothing about where any particular decision inside that business is settled, nothing about who is asked first, and nothing about how long anything takes. No price for that business is published anywhere, no cost and no accounts of any kind, so there is nothing to work the four questions against even indirectly.

Now the board fact, and here the reading has to be careful because the temptation runs one way. Appointing the majority of a board is more or less what deciding how a company is run consists of. So that appointment settles that one company can decide how another is run. The appointment is genuine control, and it is published. The very next sentences of the same disclosure state that nothing was merged, that two boards still sit, that two bank accounts still run and that two sets of statements are still prepared. Chitra Binding Works Private Limited did not stop being a company with its own disclosureSomething a business states in a document it publishes, as against something a reader worked out from what it stated. obligations the day it became a subsidiaryA company in which another company holds a large enough share to decide how it is run..

Being able to decide and actually deciding are two different facts, the first is published and the second is not. Nothing anywhere says whether a single question inside the smaller company travels to the larger one before it is settled, or whether it is settled where it always was. The published sentence establishes a power. The published sentence records no use of that power, on any decision, on any day.

The fuller version of that argument, including how somebody outside might try to tell the two apart, is covered separately under Centralised vs Decentralised Organisation. The distinction is where this guide stops, and nothing carries across.

What would have to be disclosed before there could be a third fact?

A refusal can be turned into something that can be handed to somebody. Saying that a subject is not visible from outside, and leaving it there, costs nothing and teaches nothing. The useful move is to write down exactly what would have to be published, item by item. Then the emptiness has a shape, and somebody could in principle go and ask for it.

Six things, worth reading slowly. Each one closes off a step the next one needs.

Now land it. Not one of the six is published for any business in these notes, and the reason is not that anybody withheld them. A public filing regime asks a business to put its accounts on record. The regime does not ask which questions travel and which do not, and almost no business volunteers any of the six anywhere else either. So the six blanks describe what the regime asks for rather than describing anybody's conduct.

A card with six empty rows and a written reason beside each one is a finding, and a diagram with six full boxes drawn from nothing is not. The first can be handed to somebody who might supply a row. The second cannot be argued with at all. There is nothing in it to disagree with except the confidence of whoever drew it.

Transformation: what does changing a structure actually cost?

Transformation usually carries a sense that cannot be used here. The word names a business rearranging itself on purpose: work moving, decisions moving with it, somebody deciding that questions which used to travel should stop travelling. Nothing published anywhere in these notes says that either business did anything of the kind, so no rearrangement can be described and none can be claimed. The one movement that was published can still be priced, and the limit of what that price shows can be stated exactly.

Here is the movement. Over one published year Anjani Stationers' standing baseThe spending that arrives in the same size whichever way volume goes, so it is earned back out of volume every year. went from Rs 49,60,000/- to Rs 74,00,000/-, a rise of Rs 24,40,000/-. The disclosure of that movement names what it went on in six words: people, space and a binding operation the business bought into. Not one of those three carries a price. Six words of purpose sit on top of one number, and that is what most readers outside a business ever get.

A breakdown of that rise does exist, and it is worth being exact about what kind of object it is. One is supplied elsewhere in these notes, and its caption stamps it in its own words: the split is an estimate, not a disclosure. Its body sentence explains why the estimate is honest rather than invented: the components are on the face of the statement even though the split is not. So somebody read three reported lines and worked out how much of each was new, and then said out loud that this is what they had done.

The split runs: employee benefitsWhat a business spends on the people it employs, taken together as one reported line. Rs 6,00,000/-; the fixed part of other operating costs Rs 11,40,000/-, mostly the second warehouse taken during the year; depreciationThe spreading of the cost of a long-lived asset across the periods that use it. and amortisation Rs 7,00,000/- on the assets bought. The three add to Rs 24,40,000/- exactly. And the stamp travels with every one of them, every single time: the split is an estimate, not a disclosure.

Now work the two limbs this section actually needs, and notice how quickly each of them runs out. The people limb, the estimated rise in employee benefits of Rs 6,00,000/-, is the closest these notes come to putting a price on changing how a business is arranged, and the split it belongs to is an estimate, not a disclosure. The people limb still does not say how many people, in which part of the business, doing what, or whether a single person's job changed at all. A business can spend that on paying the people it already had.

The space limb, the estimated rise in the fixed part of other operating costs of Rs 11,40,000/-, mostly the second warehouse taken during the year, and again an estimate rather than a disclosure, is the closest they come to pricing where the work happens. The space limb still does not say what moved into that warehouse, who decides what goes there, or whether anything that used to be settled in one place is now settled in another.

The third limb stands apart. The estimated Rs 7,00,000/- is depreciation and amortisation on assets that were bought, it is part of the same estimate rather than a disclosure, and it is never read as the price of the binding operation the business bought into. Paying cash for a shareholding in another company creates no depreciation in a business's own accounts, so joining the two would be a step nobody has taken.

The estimate is the honest ceiling on what anybody outside can say, and it is worth more than a diagram nobody published. A figure that closes to the rupee and states what kind of figure it is supplies a real edge and a real limit at the same moment. Exactly where such a figure stops is visible, so it is a better instrument than a picture of an arrangement.

Hold it at street level for a second. A stall outside one office takes a second pitch on a yearly rent. The second pitch is standing there in the open, so everybody walking past can see the rent. Whether anybody's job changed, whether the person who used to decide the day's order still decides it, whether one pitch now waits on the other: none of that is visible from the pavement on day one, or on day four hundred.

Would a diagram have answered the question anyway?

Here is the uncomfortable part, and it is the part that makes the refusal worth reading rather than worth apologising for. Suppose one of these businesses had published a diagram of its arrangement. The diagram would have answered a different question from the one the reader arrived with.

A drawing of an arrangement records what somebody meant to set up. The question was where a decision actually gets settled, who has to be asked first and how long it takes. The intended arrangement and the record of settled decisions are two different documents, and only the second answers the question asked. The claim sounds like a debating point until the behaviour of each document is examined, so it has to be earned rather than asserted.

Take permanence first. An intended arrangement is set out once, on a day, and then it sits there. The places where decisions actually get settled move continually: somebody leaves, somebody proves reliable, one question gets settled twice by the same person and then always gets settled by that person. The drawing does not move with any of that. So the two begin drifting apart from the day the drawing is finished, and nothing about the drawing says how far apart they have got.

Now take what gets recorded. Somebody happens to know an answer, everybody learns that, and a question starts going round them informally. A drawing cannot contain a question of that kind. Informal routes like that are not a small residue. Such a route often decides how fast a business can move at all, and it stays invisible to the instrument that was supposed to show the arrangement.

The question is not what the arrangement looks like, it is where the next decision will be settled and how long that will take. An invented arrangement would have been easy to draw, and the ease is the problem. The drawing would answer a question nobody asked, and it would be believed.