Vertical vs Horizontal Integration: Which Way to Grow
Vertical integration moves along the chain and takes over a stage the business does not run. Horizontal integration moves across it and buys more of the stage it already runs. The two answer different questions, so neither is generally better. At Anjani Stationers the direction is settled by one fact: binding set the pace, and its own lines already run at 62.50 per cent.
Both of those sentences rest on ground already laid. Growing along the chain is set out in full under Vertical Integration: Owning More of the Chain, where growing sideways is named as the thing it was not. Neither direction is built again here. The comparison below puts the two side by side, works out when each of them is the wrong move, and tests the whole argument against a rival business drawn to make the test hard.
Anjani Stationers Private Limited, an invented manufacturer of hard bound registers, carries the figures used all through these notes. In the year already published it made and sold 2,50,000 registers. Its cutting line runs 150 registers an hour, its printing line 125 and its binding line 100, and because a works can only move at the pace of its slowest step, that 100 an hour across two lines, eight hours a day, 250 days a year, sets a rated capacityThe most a works could turn out in a period if every line ran its full hours at its normal speed, before anything real interrupts it. of 4,00,000 registers. Making 2,50,000 against a rated 4,00,000 is 62.50 per cent. Hold that fraction. Almost everything below turns on it.
Which question is this comparison actually asking?
There is a habit worth breaking first. A reader who meets one comparison after another in a single stretch of reading tends to file them all as the same shape of thing, two options, pick one. The comparisons are not the same shape. Each one is a different question with a different unit of decision, and mixing them up is how a discussion about growth quietly turns into a discussion about stock levels without anybody noticing the change of subject.
A comparison is only useful once the question it answers is known, and this comparison asks which direction a business should grow in. The question starts after a business has already decided that it wants to be bigger. It does not ask whether growing is wise, what any of it would cost, or what anybody is worth. The question is narrower: given that more is wanted, does the business reach along the work or across it?
| The question on the table | Where it is settled | What comes out of it |
|---|---|---|
| Given that the business wants to grow, which direction does it grow in? | This guide | Along the chain, or across it, and when each is the wrong choice |
| Is this one step of the work worth having inside the business at all? | Make vs Buy: Whether a Step Is Worth Owning at All | One step, judged at one volume, against what an outside supplier charges |
| How much paper should be sitting in the shed? | Just-in-Time vs Just-in-Case Inventory: How Much Cover to Hold | Days of cover, and the length of stall at which holding stock pays |
In the middle column the difference stops being subtle. Because a supplier will quote for the work by return of post, the second question can be answered without buying anything at all. The third question is answered in working days and never in machines. Only the first one is about the shape of the business getting larger, and only the first one is the subject here.
What separates the two directions, in one line each?
Along the chain means taking over a stage of the work the business does not run. Across the chain means buying more of a stage it already runs. The distinction is no bigger than that, and it needs no more room than this.
The only test that separates the two directions is whether the business already runs the stage in question, and no other feature of the deal changes the answer. Not the size of the stake, not whether the two businesses sell into the same trade, not whether the seller is willing. A tea stall never kept cows, so taking over the dairy that supplies its milk moves it along its chain. Pouring tea is precisely what the stall already does all day, so taking over the tea stall on the next corner moves it across.
Anjani Stationers takes over the paper mill that supplies it. Which direction of growth is that?
Why are these not two answers to one question?
Here is where most of the confusion lives, so take the next few lines slowly. The two directions are usually presented as rival proposals, as though a works could sensibly line them up and pick the cheaper. The two directions are not rivals. They are answers to two different questions that happen to arrive on the same desk in the same week.
Growing along the chain answers this: should a step of the work be inside the business rather than bought from somebody else? Growing across the chain answers a different one: should the business be bigger at a step it already performs? A works can want both at once, want neither, or want one urgently while the other would be actively harmful. None of those four positions is a contradiction. The two questions are independent.
Choosing between the two directions starts by working out which question is actually being asked, and a business that has skipped that step will find both proposals persuasive at the same time. Finding both persuasive is not a failure of judgement. Each proposal is a good answer to something, so the double persuasion is exactly what should be expected. The trouble is that only one of the somethings is the problem the business has.
A works has decided to grow and two proposals arrive, one to take over a supplier and one to buy a rival. Why do both sound persuasive?
When is growing along the chain the wrong move?
Taking over a stage means taking over its rate. The transfer of rate is the whole mechanism, and it means the move is only worth making if the rate taken over was the one holding everything else back. Anjani Stationers runs its three lines at three different speeds, so it shows both halves of that sentence.
Cutting could get through 6,00,000 registers a year if it never waited for anything. Printing could get through 5,00,000. Binding, at 100 an hour, tops out at 4,00,000, and since a register has to pass through all three, the works finishes 4,00,000 at best. Now suppose somebody proposes taking over the cutting operation and doubling its speed to 200 an hour. Cutting rises to 8,00,000 a year of theoretical throughput. The works still finishes 4,00,000. Binding has not moved. Do the same to printing and the answer is 4,00,000 again.
Taking over a stage that already has spare rate moves nothing at all, however much the stage improves in itself. The idea that a sequence moves at the speed of its slowest step belongs to Eliyahu Goldratt, who put it at the centre of his 1984 book The Goal, and it is the one idea worth carrying into any meeting where somebody proposes buying a supplier. Anjani's own purchase passes that test squarely: at the start of year two it took 70 per cent of Chitra Binding Works Private Limited for Rs 21,00,000/-, and binding is precisely the stage that had been setting the rate. Lift binding to 125 an hour and rated capacity rises to 5,00,000. Nothing else in the works can say that.
There is a second and quieter case where the move along the chain is wrong, and it needs only a line. If the stage can be bought reliably, at a rate that holds, from more than one supplier competing for the work, then having it inside buys very little that a contract did not already give. The purchase does buy a set of machines to keep busy, staff to keep paid and a step to keep running on the days the business would rather have been doing something else.
Anjani Stationers took 70 per cent of Chitra Binding Works Private Limited, its binder, for Rs 21,00,000/- at the start of year two. Cutting runs 150 registers an hour, printing 125 and binding 100. What made binding the stage worth taking over?
Anjani Stationers made 2,50,000 registers against a rated capacity of 4,00,000. What would buying a rival's machines add?
When is growing across the chain the wrong move?
The mirror image, and it is easier to feel than to state. Growing across the chain is wrong when the business is already holding unused capacity at the very stage it proposes to buy more of.
Think of a household with a car that sits in the compound six days out of seven. Nobody in that household is short of a car. The household may be short of somewhere to go, short of time, short of a driver, short of petrol money. Buying a second car answers none of those and adds an insurance premium. The shortage is real, and a second car is not what is short.
Anjani Stationers is that household. Its 4,000 line-hoursOne hour of running on one production line. A works with two lines going for eight hours has put in sixteen line-hours that day. a year at the binding rate of 100 could finish 4,00,000 registers. The 2,50,000 it actually made used 2,500 of those hours. The remaining 1,500 line-hours a year are hours in which the machines were switched off and the fixed costs carried on anyway. In register terms that is 1,50,000 registers of capacity nobody used.
Idle capacity cannot be fixed by buying more capacity, and a works running at 62.50 per cent that acquires a second works has bought itself a second way of running at 62.50 per cent. Notice the limit of that claim. The second works is not worthless. The part of it made of machines is answering a question this particular business did not ask.
What would buying a same size rival actually add?
Now the test. Bhavani Register Works is a 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. maker of registers invented for this comparison, and it makes 1,50,000 registers a year. Anjani Stationers has 1,50,000 registers of capacity standing idle. The two figures are the same number.
The symmetry is a teaching device and not a discovery. Bhavani was drawn at exactly the size of Anjani's idle capacity. The machines on both sides then cancel to nothing, and whatever is left standing afterwards is the only thing genuinely being bought. Real rivals do not line up this neatly. Bhavani was built to.
Follow the consequence. Anjani made 2,50,000 registers and Bhavani makes 1,50,000. Together that is 4,00,000, Anjani's own rated capacity to the register. Every stage can take it: cutting has 3,50,000 registers of headroom, printing has 2,50,000, and binding has exactly 1,50,000. Binding's headroom is the whole of Bhavani's output and not one register more. Making those 1,50,000 registers would need 1,000 hours of cutting, 1,200 of printing and 1,500 of binding, and 1,500 idle line-hours a year is precisely what Anjani has.
Bhavani Register Works makes 1,50,000 registers a year and Anjani's idle capacity is 1,50,000 registers. What follows from that being the same number?
If the machines cancel, what is left to buy?
Strip out everything that cancels and look at what is standing. Bhavani's cutting machines, printing machines and binding machines all have counterparts at Anjani that are switched off for 1,500 hours a year. Its shed has a counterpart. Its capacity has a counterpart. Every register it makes is a register Anjani could make without buying anything from anybody.
One thing has no counterpart. The shops and schools that place orders with Bhavani place them with Bhavani. A customer is not a thing a works can manufacture on a night shift, so no amount of idle machinery at Anjani produces a single one of them. The order book is the one item on the list that cannot be reproduced by switching a line back on.
The only part of this purchase actually worth having is the order book, and a buyer who cannot separate the customers from the machines will pay for both and receive one. What an order book is worth, how a figure would be put on it, and whether those customers would stay once the name over the door changed are all questions about value, covered separately. The list of what is on offer is settled here; what to pay for it is not.
If the machines cancel, what is actually being bought?
The mistake: counting the rival's output as new production
Here is how this goes wrong in practice, and it goes wrong quietly. A buyer sits down with Bhavani's figures, sees 1,50,000 registers a year, and writes them down as 1,50,000 registers the group would gain. Because those registers were already available, the arithmetic that follows is then perfectly correct and completely wrong.
Work it through in the open. Anjani's rated capacity is 4,00,000 registers and it made 2,50,000. The difference, 1,50,000 registers, is production Anjani could add on machines it already has and already pays for, in hours it is already standing idle for. Bhavani's 1,50,000 registers are the same 1,50,000 registers seen from a different side of the table. The registers get counted twice, once in the rival's output and once inside the buyer's own unused capacity, and the second count is invisible because nobody prints idle capacity on a statement.
In plain terms: the buyer pays for a set of machines and a set of customers, receives the customers, and now has two works to keep busy where one was already half empty. The word most often used to justify the arithmetic is synergyThe word a buyer uses for extra worth a purchase is supposed to release beyond what the two businesses achieve standing apart. Synergy is asserted far more often than it is demonstrated., and here it would be pointing at capacity that was free.
The remedy is a habit rather than a warning, and it takes one afternoon. Before anything a rival produces is counted, what the plant standing idle could already produce is subtracted from it. Whatever survives that subtraction is the only thing being bought.
What would change this reading?
Everything above is a conclusion about Anjani Stationers in one particular year, and it would be dishonest to let it harden into a conclusion about a direction of growth. So change one number and watch the argument turn over.
Suppose the works were running at its full rated 4,00,000 registers instead of at 62.50 per cent. Now there is no idle capacity for Bhavani's machines to cancel against. The same purchase would bring 1,50,000 registers of capacity that Anjani genuinely does not have and cannot conjure by switching anything on. The alternative would be a greenfieldBuilding fresh capacity on empty ground rather than taking over capacity somebody else has already built and is already running. build, and a greenfield build takes time nobody has during a school supply season. The machines stop being duplicates and start being the point.
The answer here is a fact about how much of Anjani's works was being used, and not a fact about the direction of growth itself. No ranking follows. Growing across the chain is neither the weak option nor the lazy one. Growing across was the wrong option for this works in this year because of a number that could be different next year, and a reader who carries away a ranking instead of the test has taken the one thing that is not true.
Suppose Anjani were running at its full rated 4,00,000 registers. Does the argument still hold?
Which question does this guide answer, and which does it hand on?
What does a buyer check before anything else?
There is an order to these two questions that almost nobody follows, and following it costs an afternoon. A count of what the business already has too much of comes before any list of what it would like to acquire. Not after the proposal arrives. Before.
The stocktake of what is already standing idle is the cheapest work in the whole decision, and the most often left until a proposal is already on the table. The stocktake needs no adviser, no data room and no due diligenceThe examination a buyer makes of what is actually being bought before committing, going through the books, the plant and the contracts item by item. team. It needs somebody to walk the works with a clipboard and write down, stage by stage, how many hours each line ran and how many it could have run. At Anjani Stationers that walk produces one line of arithmetic: 4,000 line-hours available, 2,500 used, 1,500 idle. Everything here follows from those three numbers.
The same walk has a household version, worth doing because it teaches the reflex. A wedding caterer who turns away work every Saturday and sits still on Tuesdays does not need a second kitchen. He needs Tuesday customers, a completely different purchase. A tailor with two idle machines and one busy one does not need a third machine. He needs whatever is jamming at the busy one. In both cases the shortage and the proposed purchase are in different places, and the only way to notice is to have counted first.
The reflex generalises past register works. Because a borrower buying capacity while sitting on unused capacity is buying a problem twice, a lender reading a proposal to fund an acquisition asks what the borrower's existing plant was doing all year. An analyst who sees a works at 62.50 per cent and then sees it announce a purchase of more of the same asks what the buyer believes the shortage is. And anyone running something small can ask the same question over a cup of tea: what do I already have that is not being used, and is my proposed purchase more of that, or more of what is actually short? The bottleneckWhichever step in a sequence works slowest, so that everything before it piles up waiting and everything after it sits starved of work. is where the answer usually lives, and it is rarely where the proposal is pointing.
A colleague asks which direction of growth is better in general. What is the honest answer?
Is either direction better than the other?
No, and the refusal is worth a precise reason rather than a shrug. Ranking two things requires a criterion. The two directions do not supply one between them, and they are not answering the same question. Growing along the chain is not the cautious direction and growing across it is not the ambitious one. Each one is right exactly when the condition attached to it is met, and neither condition has anything to say about the other.
Notice how quickly the ranking becomes obvious once a question is named. If the question is whether the rate of the works can be raised, only the stage that sets the rate can answer, and that is a move along the chain. If the question is whether a works with every line full can serve more buyers next year, only more capacity can answer, and that is a move across it. Neither of those sentences ranks a direction. Each one names a condition and reads off the answer.
And the counterfactualA situation written out deliberately so that an argument can be tested against it, describing something that did not happen rather than something that did. here was built symmetrical to make exactly that point. If Bhavani had been drawn at 3,00,000 registers, half of it would have been useful capacity and half of it duplicate, the argument would have come out muddy, and a reader would have been left arguing about the split rather than seeing the principle. Drawn at 1,50,000, the machines go to nothing and the principle stands out on its own. A well built comparison exists to make a principle stand out like that.
What a legal form here does and does not settle
The legal form used here is the Indian private limited form, and it appears only to say which set of accounts a figure belongs to. Anjani Stationers Private Limited and Chitra Binding Works Private Limited carry it. Bhavani Register Works carries none. How a holding of one company in another is presented, and what has to be filed when one business takes over part or all of another, are matters of company law and of accounting standards. The current position should be read at its own source before anything is relied on.
Where does any of this come from?
No outside authority settles which way a works should grow. The direction is decided by the plant a business already runs, a fact about that business rather than a rule somebody published. One book is named below because it gave the language of the stage that sets the pace. Everything else is arithmetic on figures belonging to businesses nobody can visit.
| What is named | Why it is named here | Where to read it | Checked on |
|---|---|---|---|
| Goldratt, E. M., The Goal, 1984 | Gave the idea of the step that sets the pace for a whole sequence, which is why binding rather than printing decides what this works can finish | A printed book, so no site is given | 23 August 2026 |
| Ministry of Corporate Affairs | Where the filed accounts and the shareholding of an Indian manufacturer of this kind would be found | mca.gov.in | 23 August 2026 |
| The two subjects this one stands on, Vertical Integration: Owning More of the Chain and Throughput: The Rate the System Actually Produces | Carry the two definitions and the three line rates that this guide assumes rather than rebuilds | Elsewhere in these notes | 23 August 2026 |
Anjani Stationers Private Limited, Chitra Binding Works Private Limited and Bhavani Register Works are invented.
Educational material. Not advice on any investment, tax, budget or market position.