The Operating Model: How the Work Actually Gets Done

An operating model is the answer to one question: what has to happen for one unit to reach one buyer. Anjani Stationers converts paper into registers, so its answer is a sequence of physical steps. Setu Bazaar converts nothing, so its answer is a sequence of handoffs. The two answers share not one step. To analyse either, list the steps, price each one per unit, and find the slowest.

Underneath that sits a decision almost nobody makes out loud. Before a single rupee can be put against anything, somebody has to decide what one unit is, and no financial statement will decide it. A statement reports what a whole year of business cost. A statement has no idea what one register cost. A register is not a category any statement knows. Anjani Stationers Private Limited published its rupee totals for that year and nothing underneath them. The physical layer beneath those rupees, meaning the count of registers, the price one register fetched, and the cost of each step, was published nowhere. Each per-register figure below earns its place for exactly one reason: it multiplies back to a published total precisely, to the paisa. Rebuilding a published total is the method.

What question does an operating model actually answer?

Start with a tailor, a shop almost everybody has stood in. A length of cloth comes in the door. Somebody measures the customer, cuts the cloth, stitches the pieces together, and hands over one shirt. Four things happened, and every one of them was work a person had to perform. Next door is a shop that buys finished shirts by the dozen and sells them one at a time. A customer walks out of either place holding one shirt. Beyond that, the two shops have almost nothing in common at all.

An operating model is a sequence of steps, not a description of a business. An operating model is not the story a business tells about itself, not the market it sells into, and not how much money it made. The model is the list of things that have to happen, in order, before one buyer is holding one unit. Change that list and the business has changed in the only way that shows up on a factory floor. Two businesses can sell the same object to the same person and run entirely different lists, and the rest of this guide turns that observation into a procedure that can be carried out on a company nobody has ever seen inside.

Where does the analysis start, and why is it always the unit?

Anjani Stationers Private Limited makes hard-bound registers of one hundred sheets, and one register is its unit. One hundred sheets printed on both sides give a register its two hundred numbered sides. Paper reaches the works in a reamA ream is a standard bundle of paper. Anjani Stationers buys its paper in reams of five hundred sheets.. A ream holds five hundred sheets, so one ream is five registers if not a single sheet is wasted anywhere. The ream arithmetic describes the shape of the unit and nothing beyond it. How many registers actually came out of the reams the worksThe works is the factory floor and everything standing on it, as against the office that sells what the floor makes. consumed, and how much paper was lost on the way, is a separate argument taught on its own.

Setu Bazaar's unit is one delivered order. Not one buyer, not one seller, not one rupee of goods moving across it. One order, placed by somebody, delivered to somebody. Naming that unit is what makes the rest of the analysis possible, and it is also where an analysis most often goes quietly wrong. A business will happily describe itself in terms of buyers when its costs are actually driven by orders.

Without a named unit, not one step can be priced, and every operations argument that goes wrong goes wrong here first. A household knows this instinctively. The total a wedding cost means nothing until somebody asks how many guests ate. Four lakh rupees for two hundred guests and four lakh rupees for six hundred guests are two completely different pieces of news, and the difference only appears once the guest has been named as the unit. Businesses lose this discipline constantly, and the loss is invisible until somebody tries to price something.

What are the steps, and do two businesses ever share them?

Write both lists down, side by side, and mark each step according to whether the business does it itself or has it bought inA step is bought in when somebody outside the business performs it and is paid for it, as against a step performed inside the business by its own people and machines. from somebody outside. Anjani Stationers: paper arrives in reams from a merchant, sheets are cut to size, cut sheets are printed, printed sheets are bound into registers, registers are packed, and packs are carried to the buyer. Six steps. Three of them are done inside and three are bought from somebody else. The binding is bought from Chitra Binding Works Private Limited, a separate company doing that one step.

Setu Bazaar: a buyer places an order, the order is routed to a seller, the buyer pays, a courier delivers it, and sometimes it comes back. Five steps. Nothing on that list changes the shape of anything. The item that reaches the buyer is the same item the seller had, and every step is a message, a payment or a movement.

The two lists share not one step, and that is the whole meaning of saying two businesses have different operating models. One list contains the word cut and the other does not. The one word cut, present on one list and absent from the other, is what everything below turns on, and it carries more than the word itself suggests.

How is a rupee put against every step, and how is it checked?

Anjani Stationers sold 2,50,000 registers in the year at a realised priceThe realised price is what buyers actually paid on average, after every discount and allowance, as against the price printed on a list. of Rs 108.00/- each. Multiply those two and Rs 2,70,00,000/- comes out, precisely the revenue the business published. Paper works out at Rs 59.40/- a register, and 2,50,000 multiplied by Rs 59.40/- is Rs 1,48,50,000/-, exactly the published cost of materials consumedCost of materials consumed is the accounting line for the value of raw material actually used up during the year, as against the value of material bought during it.. CarriageCarriage is the cost of physically moving goods, paid to whoever does the moving. Here it is grouped with packing on a single published line. and packing works out at Rs 2.40/- a register, and 2,50,000 multiplied by Rs 2.40/- is Rs 6,00,000/-, exactly the published figure. Add the last two and Rs 61.80/- of cost moves with each register.

A per-unit figure that fails to rebuild the published total is wrong, and the multiplication is the only check available. Nobody audited Rs 59.40/-. Nobody disclosed it. Somebody worked Rs 59.40/- out from a published total, and every other per-register figure below was worked out the same way. The single reason a reader should extend any credit to them is that all of them rebuild something a business already put its name to. When one of them does not, something valuable has been learned rather than nothing: either the unit count is wrong or the per-unit figure is wrong, and working out which of the two it is turns out to be the whole exercise.

A tea stall makes the same point in three numbers. If a stall charges Rs 10.00/- a glass and reports Rs 30,000/- for the month, it is claiming three thousand glasses. If the pot and the hours cannot produce three thousand glasses, then one of those two numbers is wrong and the reporting has to stop until it is settled which.

Which costs move with the register, and which do not?

Rs 74,00,000/- of Anjani Stationers' cost does not move with the number of registers. The building stands whether the machines run or not. The people who maintain them are there for the year rather than for the register. Spread that Rs 74,00,000/- across 2,50,000 registers and it comes to Rs 29.60/- each. Take the Rs 61.80/- that moves and the Rs 29.60/- that does not off the Rs 108.00/- realised price and Rs 16.60/- is left on a register, and 2,50,000 multiplied by Rs 16.60/- is Rs 41,50,000/-, exactly the published operating result. Every line in that build rebuilds something.

Rs 29.60/- is not a property of a register, it is Rs 74,00,000/- divided by however many registers there happened to be. Cut one register open and there is Rs 59.40/- of paper in it. There is no Rs 29.60/- of anything. The Rs 29.60/- was never in the register. The figure is produced by a division, and the divisor is a fact about the year rather than a fact about the object. The distinction between a cost and a share of a block sounds pedantic right up to the moment somebody quotes a price off it. Somebody does exactly that below.

One caveat carries the whole build. Which of these rupees follow the register and which stand still is a judgement somebody formed while reading the accounts. The sorting is not something any statement sets out. Nowhere does a published line say which rupees rise and fall with the count. A person looked at the accounts, decided, and wrote the sorting down, and every argument standing on that sorting inherits the judgement. The judgement is worth naming each time it is leant on.

Why does one cost per register fall while another cost per order does not?

Anjani Stationers' total cost per register falls as the number of registers rises. The fall happens for one reason, and the reason is not efficiency: Rs 74,00,000/- of the cost does not move, and a block that does not move is being shared among more registers each time. The Rs 61.80/- that moves with the register does not fall at all, at any volume, ever. The same Rs 61.80/- sits on the first register and on the two hundred and fifty thousandth. So when the total cost per register falls, every paisa of the fall came out of one part of it.

Setu Bazaar's cost of revenueCost of revenue is the accounting line for what the thing sold actually cost the seller, before any of the costs of running the business are counted. is Rs 120.00/- an order, and that figure does not fall at all. Nothing is being converted, so there is nothing to spread. The cost is the goods themselves, charged again on every single order, and a five hundred thousandth order costs Setu Bazaar what the first one did. Across 5,00,000 orders Setu Bazaar's total variable cost is Rs 200.00/- an order, and it moved Rs 500 crore of goods to keep Rs 20 crore of revenue. Its result for the year was minus Rs 2.5 crore.

Conversion is what creates a block of cost that does not move, and a block of cost that does not move is what makes volume worth anything. A shed, a cutting machine, a printing line and the people who keep all three running are the price of turning one thing into a different thing. Bought once, they are bought for a year regardless of how much they are used. A business that converts nothing never has to buy that block. Never buying the block sounds like an advantage right up to the moment somebody asks what more volume is supposed to do for it.

Both claims have to be said in full. Dropping half of either one turns a true sentence into a false one. Anjani Stationers' figure that falls is its total cost per register, not the part that moves with the register. Setu Bazaar's figure that stays flat is its cost of revenue per order, not every cost it carries. Written loosely in either direction, the sentence stops being true and starts sounding like a rule about businesses in general. No such rule holds.

Which step sets the pace for all the others?

Consider a queue of four counters where the third counter is slow. The queue is not really a queue of four counters. The queue is one slow counter and three people watching, and putting a faster clerk on counter two gets nobody out of the building any sooner. A chain of steps runs at the speed of its slowest step, and speeding up any other step changes nothing at all until the slowest one changes. Eliyahu Goldratt set this out in The Goal in 1984, and the name belongs with the claim.

The slowest step sets the rate for every step around it, and every other step's speed is spare until it moves. At Anjani Stationers the slowest of the six steps is binding. Naming binding is worth doing before any arithmetic touches it. The name alone shows where an improvement would land and where an improvement would evaporate. Money spent making the printing faster buys a taller pile of printed sheets sitting in front of the binding table. Nothing more reaches a buyer.

How fast each step actually runs, and what the works could therefore finish across a whole year, is a separate argument taught on its own, and so is turning that rate into a figure for how much of the works was used. Naming the slowest step is as far as a description alone can carry the analysis. Everything after it needs measurement.

How to Analyse a Company’s Operating Model

Six steps, in this order, each one naming what gets written down and what gets checked. Every argument above was the reasoning behind one of the six.

1. Name the unit, then name what one of them fetched. Here the unit is a single hard-bound register of one hundred sheets and the figure is a realised Rs 108.00/-. Write the unit down in physical words rather than in money. CheckRevenue divided by the price gives a unit count, and that count has to be a number the business could physically have produced. If it comes out at a number no shed could make, the price is wrong.
2. Write out every step from raw input to buyer, and tag each one bought in or performed inside. Six for Anjani Stationers, five for Setu Bazaar, and no overlap between them. CheckAn untagged step is one that has not yet been understood. Where who performs it and who is paid for it cannot be stated, that has to be found out before anything is priced.
3. Put a rupee against each step, per unit. Paper at Rs 59.40/- a register. Carriage and packing at Rs 2.40/-. And so on down the whole list. CheckThe rupees must come from something published. Where one has had to be estimated, the word estimate goes beside it and stays there. An estimated figure and a figure the business disclosed do not deserve equal weight.
4. Multiply every per-unit figure back by the unit count. Rs 59.40/- multiplied by 2,50,000 is Rs 1,48,50,000/-, a total somebody already published. CheckIf a line does not rebuild its published total to the rupee, either the unit count or the per-unit figure is wrong. Finding out which of the two it is, rather than adjusting one until they agree, is the whole exercise.
5. Sort the costs by whether they move with the unit, and record that the sort is a judgement. Rs 61.80/- moves with a register and Rs 74,00,000/- does not. CheckName the one cost that was least certain. The least certain cost is where a later argument will break, and naming it now costs nothing while naming it later costs an argument.
6. Find the slowest step. At Anjani Stationers that is binding, and everything else has spare speed. CheckAsk what would have to change for that step to stop being the slowest, and whether the business controls that thing or somebody else does. The answer changes what an improvement is worth.

Four of the six steps need nothing the company did not already publish, and the other two need a judgement that has to be labelled as a judgement. Steps three and five are where a person enters the arithmetic. Everything else is reading and multiplying. Four out of six is worth noticing. An operating model can be taken apart from the outside far further than most readers expect, and the two places where it cannot are the two places where every disagreement will eventually be found.

What does describing an operating model not settle?

Whether the business earns anything is not a question an operating model answers. A description does not say whether the way of working is a good one, whether anybody else could copy it, or what any of it is worth to anybody. A description states what has to happen and roughly what each happening costs, and it stops there.

An operating model is a description and not a verdict. The temptation to convert one into the other is strongest exactly where two businesses stand next to each other. Setu Bazaar has no conversion, no slowest step in a shed and no pile of paper to run down, and the reason is not that it is the better business. The reason is that it makes nothing. Setu Bazaar's own result for the year was minus Rs 2.5 crore and Anjani Stationers' was a positive Rs 41,50,000/-. Two facts do not amount to a ranking. The absence of an operations problem is not the presence of a better business.

How a lender, an analyst and a household actually use this

A lender looking at a manufacturer's request for a larger limit is not really asking about the profit. The lender is asking which of the borrower’s costs will grow with the extra volume the loan is meant to fund and which will not. The second group is where the repayment is going to come from. A borrower whose costs are almost entirely a rate charged again on every unit has very little to gain from being bigger, and a lender who has read the step list knows that before the projections arrive.

An analyst uses the multiply-back in reverse. Given a published revenue and a plausible price per unit, the implied unit count either is or is not a number that shed could produce. When the implied count is impossible, one of the two inputs is wrong, and the analyst has found a question worth asking rather than a number worth reporting. A question worth asking is a far more useful outcome than a tidy per-unit figure nobody can check.

A household does the same arithmetic without naming it. A tiffin service run out of a kitchen has a gas connection, a set of vessels and one person's time, and none of those grow with the fortieth box. The rice and the vegetables do. Anybody who has worked out that forty boxes a day pays and fifteen does not has just sorted their costs by whether they move with the unit, and has just discovered why the block matters.

What goes wrong when a share of a block is quoted as a cost?

The error that matters most is made by careful people holding a correct arithmetic sheet.