The Post-Merger Review: Testing the Rationale Against the Outcome

A post-merger review tests the argument made before a purchase against what actually happened: what was promised, what arrived, what it cost, and which of the original assumptions turned out to be carrying the weight. A review triggered only by disappointment can never tell a sound decision apart from a lucky one, so the review runs on schedule whether the outcome looked good or bad.

Underneath that sits one uncomfortable idea. A decision and a result are two different objects. The first is chosen with the information available on the day. The second arrives later, shaped partly by the choice and partly by everything nobody could see. A review that reads only the second and reports on the first has quietly swapped one for the other, and the swap is invisible in the finished document. So the work of a review is mostly separating those two, keeping them separate in the finished document, and being honest about where the evidence runs out.

What purchase is this review being run on?

One invented transaction carries every figure below. Every number in it can be followed all the way through rather than asserted. The buyer is Harivansh Packaging Limited, an invented maker of rigid and flexible packaging for the food and personal care trades, listed on both Indian exchanges. Harivansh Packaging took every share in Sundarban Polymers Private Limited, an unlisted business whose product is flexible film. The target's earnings before interest, tax, depreciation and amortisation (EBITDA) ran at Rs 132 crore. Ten times that EBITDASpelled out, that is earnings before interest, tax, depreciation and amortisation, so it reads as trading profit taken before borrowing costs and before any charge for money spent in earlier years. set an enterprise valueWhat a whole business is being valued at once the lenders' money and the shareholders' money are both counted in. It is built up in the valuation notes and used here as a finished tool. of Rs 1,320 crore. Strip out the Rs 180 crore of net debtWhat is left of a company's borrowings once the cash it holds is set against them. It counts here because on a full purchase the target's borrowings come across rather than vanishing. that Sundarban Polymers already carried, and what reaches the sellers for their shares is an equity valueThe sum handed to the sellers for their shares once whatever the target itself had borrowed is removed from the value of the whole business. Routinely muddled with that larger figure, and never equal to it. of Rs 1,140 crore. The figure a review has to work on is the Rs 1,140 crore, not the Rs 1,320 crore.

The buyer met that price two ways. Rs 140 crore came out of cash the buyer already held, and the remaining Rs 1,000 crore came from fresh debt drawn at the 9.0 per cent written into its own borrowing contract. On completion day the buyer's earnings per shareTake the profit after tax and spread it over every share in issue. Because the reading is per share, it shifts when the profit changes and it shifts when the share count changes. landed at Rs 12.14/- where Rs 12.50/- had stood before. The fall is dilutionWhat happens when a transaction itself pushes a per-share figure down. Push it up instead and the word is accretion. Both describe arithmetic and neither passes judgement on the purchase. of 2.9 per cent. The argument written before completion accepted that fall on one condition: that the purchase would produce a further Rs 8.67 crore of EBITDA, counted after the expense of getting it, measured against a baseline fixed before the deal closed. The condition in that argument is the sentence a review comes back to test. Ashwin Rege, who leads the transaction team, drafted that sentence. Devyani Kulkarni, who is chief financial officer at Harivansh Packaging Limited, put her name to it.

When does a review happen, and who decides the date?

Most teams, asked when the review happens, answer with something shaped like a shrug: when things have settled, or when somebody asks. Both of those are triggers rather than dates, and a trigger is exactly what a review must not have. The date of a post-merger review is a design decision made before completion, not an administrative one made afterwards, because a date left to be chosen later is a date chosen by the result.

Consider a household that has just moved to a new city for a job. In the first month, asked whether the move was working, the household talks about boxes, a school admission that has not come through and a landlord who will not fix the geyser. None of that is the move. Four years later the honest answer is that the move can no longer be separated from everything else that has happened since. There is a middle window where the question is answerable, and it is not at either end.

A purchase has the same shape. Run the review too early and the changes have simply not landed: the supply contracts have not been renegotiated, the second plant has not been reconfigured, and the register that is supposed to hold the delivered figures has almost nothing in it. Run it too late and two separate kinds of evidence have decayed. The figures have been restated, reclassified and absorbed into a larger business until the acquired part cannot be traced. And the people who made the decision have moved on. Their beliefs on the day, and the reasons for them, are the one thing nobody can reconstruct.

The review date belongs in the original argument, fixed before completion, for exactly the same reason the baseline does: both are cheap to write down in advance and impossible to establish honestly afterwards. Nobody can retrofit a date that was not chosen in advance, because the choice of date is itself informed by the result. A date set early produces a review that finds nothing. A date set late produces a review that finds a story. A date set in advance produces a review that finds whatever is there.

How to run a Post-Deal Review: what are the six steps, and in what order?

A post-deal review is a comparison, and a comparison needs both sides present before it starts. Most reviews go wrong by assembling only the second side, the outcome, and then reasoning backwards to what the first side must have been. The order below exists to stop that. Retrieving the original document therefore comes before touching a single delivered number.

Step one is to retrieve the original argument exactly as it was written, resisting every temptation to tidy it. The value of the document is that it was written by people who did not know what is known now. Step two is to restate its four parts and its falsification sentence in the review itself, so the thing being tested is written down rather than held in somebody's memory. Step three is to pull the delivered figures from the register against the baselines fixed before completion, not against last year, not against a budget written afterwards. Step four is to recompute the funding cost actually incurred, usually the one part of the whole exercise where the answer is beyond argument. Step five is to compare the argument with the result part by part rather than in total. Step six is to record which assumptions turned out to be carrying the weight.

Step five is where reviews are won and lost, so it deserves its own sentence. Comparing the argument with the result in total is what allows a large miss on one part to be hidden behind a small beat on another, and the parts that offset each other are usually the two the organisation most needs to talk about. Suppose the argument promised savings from combining two purchasing operations and growth from selling the target's films to the acquirer's existing customers. Suppose the savings arrived early and larger than promised, and the growth never arrived at all. Netted together the total looks close to plan. Split apart, the review has found that the business is good at the mechanical half and has learned nothing yet about the commercial half. Somebody can act on that finding.

Step four is the odd one out, and deserves a note of its own. Every other step involves judgement about what counts and what does not. Recomputing the funding cost involves none: the borrowing was drawn at a contracted rate, the interest is arithmetic, and the tax effect is arithmetic. The funding cost is the one part of a post-deal review where the reviewer confirms a number rather than re-estimating it, and a review that produces a fresh estimate here has substituted opinion for a fact that was already fixed.

How to run a Deal Post-Mortem: what does it do that a review does not?

The difference fits in one line before the steps arrive. A post-deal review asks what happened across the whole purchase. A deal post-mortem asks what went wrong and why, on one specific failure, and it is run on that failure rather than on the transaction as a whole. The two are often confused because both look backwards, but they answer different questions and they are commissioned for different reasons.

The steps of a post-mortem run like this. Fix the specific outcome being examined, narrowly enough that everyone in the room is talking about the same event. Establish what was known at the time and, separately, what was knowable at the time. The two are rarely the same, and the gap between them is where most of the learning lives. Separate what was decided from what was assumed. An assumption that nobody noticed making is not a decision anybody can be said to have taken. Identify the point at which a different decision was actually available. Then record what would have had to be different, in evidence or in process, for that different decision to have been taken.

The fourth step is the one that decides whether a post-mortem has found anything, because a post-mortem that cannot name a point at which a different choice existed has documented a misfortune rather than an error. Documenting a misfortune is not a failure of the post-mortem. The finding is a valuable one: this particular loss is not a process problem, and tightening the process will not prevent the next one. The one thing a post-mortem must never do is manufacture a decision point that was not there, and manufacturing one is exactly what happens when the conclusion has been chosen before the work starts.

Was it a bad decision, or was it just a bad outcome?

The split between a decision and its outcome is the centre of the subject, and it repays care. Decision quality and outcome quality are independent, so all four combinations occur: a carefully made decision followed by a poor result, and a careless one followed by a good result, are both entirely ordinary events rather than contradictions.

Everybody already accepts this in a setting where the odds are visible. A driver who goes home after four drinks and arrives safely made a bad decision and got a good outcome, and nobody watching thinks the safe arrival vindicates the driving. A household that keeps six months of expenses in a deposit account and never faces a job loss made a sound decision and got an outcome that made the caution look unnecessary. The distinction is easy to hold when the process is visible and the luck is obvious. Business results hide the luck, so the distinction collapses.

Judging the quality of a decision by the quality of its result is called resulting, a term that belongs to Annie Duke, Thinking in Bets, 2018, and it is the single most common failure in every review of every past decision. The name matters because it makes the move nameable in a meeting. Once somebody can say, out loud, that the room is resulting, the conversation changes shape.

The practical consequence is the part worth carrying away. A process improves only if decisions are examined on what was knowable at the time, because results on their own cannot separate the decisions that were sound from the decisions that were lucky. A team graded on results promotes whoever drew the favourable years and demotes whoever made careful choices in unfavourable ones, and the team notices within about two cycles. The team changes not the quality of its decisions but the quality of its documentation. Documentation is much cheaper to improve, and worth nothing at all.

Case Study vs Post-Mortem: which one was written to teach something?

Both of them look backwards at a transaction and both produce a document, so they get shelved together. The two are selected for opposite things. A case study is written to teach, so it is selected for a clear story with an ending. A post-mortem is written to find out what happened in one specific instance, and it is run whether or not any story emerges.

Selection is the whole of it. Nobody sits down to write up a purchase that neither failed nor transformed anything, where the savings arrived roughly as promised, the growth mostly did not, the leverage came down on schedule and four years later the business is a little larger and about as profitable. There is no lesson with edges. So it does not get written, and it does not get taught, and it does not appear on any list of transactions worth studying. Meanwhile the collapses get written up and so do the transformations, because both make a story that finishes.

The consequence lands on the reader rather than on the transaction. Written accounts over-represent the loud outcomes at both ends, so a reader who learns transactions from case studies carries a distorted picture of what an ordinary purchase actually looks like. Such a reader is quicker to expect catastrophe and quicker to expect transformation than the range warrants, and slower to recognise the perfectly common middle where a purchase is neither and the honest answer is that the business is somewhat bigger and roughly as good. A post-mortem is not selected at all. A specific thing went wrong, and the post-mortem reports on it whether or not the account makes a story.

What does the review actually find when it is run on this purchase?

Now run the six steps on the invented transaction, in order, and on exactly the figures that were fixed before completion. The argument as written said this. Harivansh Packaging Limited hands over an equity value of Rs 1,140 crore. Rs 140 crore of that is cash the buyer already held, and the other Rs 1,000 crore is new debt drawn at the 9.0 per cent its own contract fixed. Day one earnings per share lands at Rs 12.14/- where Rs 12.50/- stood, dilution of 2.9 per cent. The falsification sentence said that Rs 8.67 crore of additional EBITDA, once the expense of obtaining it has been taken off, measured against a baseline fixed in advance, must arrive by a stated date. The falsification sentence is what a review tests.

Step four first, because the funding cost is the part beyond argument

Interest on Rs 1,000 crore at the contracted 9.0 per cent comes to Rs 90 crore. Apply the effective tax rateThe tax charge a company actually bears, read as a share of its profit before tax, which need not equal the headline rate. Here it is 25.0 per cent, and it counts because interest reduces taxable profit. of 25.0 per cent and the charge falls to Rs 67.5 crore. If the borrowing was in fact drawn at that rate, the review confirms the figure and moves on. The reviewer does not re-estimate the charge. There is nothing in it to estimate.

Against that stands the earnings the purchase bought. Sundarban Polymers contributes Rs 61 crore of profit after tax, and that figure is the record's rounded one. Worked exactly it is Rs 61.35 crore: earnings before interest and tax (EBIT) of Rs 98 crore, from which Rs 16.2 crore of interest on Sundarban Polymers' own Rs 180 crore of net debt comes off, taxed at 25.0 per cent. The rounding is load bearing and is stated plainly: Rs 61 crore is the locked figure used throughout, the exact chain would give Rs 12.16/- rather than Rs 12.14/- and a dilution of 2.7 per cent rather than 2.9 per cent, and the purchase is dilutive either way. The direction of the teaching point does not move; only the headline was slightly overstated.

Here is the part a review has to get right before anything else, and it is a discipline rather than a calculation. Both figures must be struck on the same base, because an earnings figure measured on the Rs 1,140 crore paid set against an interest cost measured on the Rs 1,000 crore borrowed produces a gap of Rs 15.95 crore that reconciles to no earnings per share figure at all. Put both on the Rs 1,140 crore and the arithmetic closes. Divide Rs 61 crore by Rs 1,140 crore and the yield reads 5.35 per cent. Divide Rs 67.5 crore by the identical Rs 1,140 crore and the cost reads 5.92 per cent. The spread is 0.57 percentage points against the purchase. On Rs 1,140 crore that spread is Rs 6.50 crore. Across 18.00 crore shares it is Rs 0.36/- and reproduces the Rs 12.14/- exactly.

So before the review has looked at a single synergy line, it can establish beyond argument that the purchase started Rs 6.50 crore, or 0.57 percentage points, behind on funding, and that this was knowable on the day it was signed rather than discovered afterwards. The sentence does a great deal of work. A business can be worth buying while starting behind on funding, so the shortfall is not a criticism of the purchase, and the argument written before completion said exactly that and set out what would have to arrive to close the gap. The sentence is instead a statement about what class of thing the shortfall belongs to. Separating what was knowable in advance from what was not is the single most valuable thing a post-merger review does, and everything else it produces is softer than this.

Then the delivery test, where this record stops

There is a gap here rather than a figure, and it is named rather than filled. The record for this transaction fixes no delivered synergy figure. There is no time series, no register extract, no reported outcome eighteen months on. An invented figure would read better and teach worse. A reader would walk away holding a number written only to keep the arithmetic tidy. So the delivery test below is taught as a method with the requirement standing in place of a result, and no figure is asserted for what actually arrived.

The arithmetic of every position can still be laid out. Whichever position an actual transaction landed on can then be run. Nothing delivered leaves earnings per share at Rs 12.14/-, exactly where day one left it. Half the requirement, Rs 4.33 crore of extra EBITDA, is Rs 3.25 crore after tax and leaves earnings per share at about Rs 12.32/-. The full Rs 8.67 crore is Rs 6.50 crore after tax. The figure is precisely the funding gap, and it returns earnings per share to Rs 12.50/-.

A review that reports delivery without stating which of those three positions it landed on has reported nothing at all. The word delivered covers the entire distance from Rs 12.14/- to Rs 12.50/-, and that distance is the whole question the argument was written to settle. The bare word delivered is the most common empty sentence in the whole class of documents, and it survives because it sounds like a finding.

The leverage check, and why the basis has to be named out loud

A review reads the balance sheet as well as the earnings, and leverage is the figure that travels furthest from the document. Leverage gets repeated in board papers, in lender conversations and in commentary, usually stripped of everything except the number. So it has to be stated with its basis attached, every time.

Read it on a consolidated basisBoth companies counted together as a single economic unit, since one of them now holds the other outright. Read them standalone instead and only the buyer's own books are in view. and the two businesses together carry Rs 1,920 crore of net debt against EBITDA of Rs 609 crore. The two together work out at 3.15 times. Inside that Rs 1,920 crore sits Sundarban Polymers Private Limited's Rs 180 crore. The Rs 180 crore travelled across with the company rather than being repaid on completion. Read it standalone and it is the buyer's Rs 1,740 crore set against the Rs 477 crore of EBITDA that Harivansh Packaging earns by itself. The standalone pairing works out at 3.65 times. Both are honest and they are not close to each other. Leverage before the purchase was 1.26 times, so either reading is a large move.

There are two honest pairings here and there is no third, so a review that divides the standalone Rs 1,740 crore by the combined Rs 609 crore has not found a third reading but a wrong one. The mixed division gives 2.86 times. The division counts one company on the top and two on the bottom, it understates the position by 0.29 turns against the consolidated reading, it belongs to no set of companies that exists, and it must never be quoted. Both of its inputs are correct figures sitting in the same pack, so the mistake is easy to make by accident.

What evidence does a review need, and what destroys it?

Three things have to exist before a review is possible, and all three are made before completion or not at all. A review needs the rationale as written, a baseline fixed before completion, and a record of what was known at the time, and the absence of any one of the three converts the review into a description of the result.

The rationale as written matters because the argument people remember is always the argument that survived. Memory edits an argument toward whatever turned out to matter, and it does so honestly and invisibly. The baseline matters because delivery is a comparison and a comparison needs a fixed point; a baseline set afterwards is set by somebody who already knows the answer, and it will drift toward whatever makes the delivered figure look reasonable. The record of what was known matters because it is the only defence against the reviewer's own hindsight. Hindsight is the strongest force in the room.

All three cost almost nothing to create in advance and cannot be produced afterwards at any price, and that asymmetry is the entire argument for creating them. Writing down the review date takes a line. Fixing the baseline takes an afternoon with the figures already to hand. Recording what was known takes a single sheet. Reconstructing any of them eighteen months later takes a great deal of work and produces something worthless, because it will have been written by people who now know how it turned out.

What can a post-merger review never conclude?

A review has a boundary, and stating it is part of the finding rather than a disclaimer bolted on at the end. Here is what a review can say on this purchase. A review can say what was promised: another Rs 8.67 crore of EBITDA, reckoned after paying for it, against a baseline fixed in advance and landing by a stated date. A review can say what the purchase cost: interest of Rs 90 crore, or Rs 67.5 crore once tax relief is counted, sitting against the Rs 61 crore the purchase bought, with both of those read on the identical Rs 1,140 crore. A review can say where leverage sits on each named basis. And with a delivered figure in hand it could say what arrived and which assumptions carried the weight.

The one thing a review cannot say is whether the purchase was a good idea. The answer to that question requires knowing what the same Rs 1,140 crore would have done somewhere else, and no review has access to the transactions that were never made. Harivansh Packaging could have spent it on capacity of its own, on a different target, on paying down borrowings, or on nothing at all. None of those alternatives was ever run, so none of them has an outcome anybody can observe. The comparison the merit question demands is a comparison against a set of things that never happened.

The boundary is not modesty and it is not a legal disclaimer. The boundary is a statement about what the evidence can carry. A review that reaches for the stronger conclusion has overreached, and it will be read as a verdict on people rather than as a finding about a decision. Nothing makes the next review useless faster. Once a review is understood as a judgement, the rationale that precedes the next purchase is written to survive one instead of to be tested by one, and the document that was supposed to make learning possible now prevents it.

Who reads the review, and what actually changes because of it?

A review with no reader and no consequence is an expensive filing. Four groups read one in practice, and each takes something different from it. The board reads it to see whether the argument put to them held together and whether the person who put it holds themselves to what they wrote. The finance team reads it for the funding arithmetic and the leverage basis, because those two travel into every subsequent conversation with lenders. The operating team reads the part by part comparison, because that is where the work sits. And whoever writes the next rationale reads all of it, which is the only reader who can change anything in advance.

The useful output of a post-merger review is almost always a change to how the next argument is written rather than a judgement about the last one, because the last purchase has already happened and the next rationale has not. If the review finds that a commercial assumption carried far more weight than anybody noticed at the time, the change is a rule that the next rationale must name its load-bearing assumption explicitly. If it finds that no baseline was fixed, the change is that the next one fixes it before signing. Each of those changes is small, cheap and permanent, and that combination is worth more than any verdict.

How does an outside reader use a review they will never see?

Almost nobody outside the company reads a post-merger review, because it is an internal document. Being internal does not make a review useless to an analyst or a lender, because the absence of one is itself readable. A credit analyst looking at Harivansh Packaging after this purchase has the public figures and can do exactly what step four does: recompute the funding cost, put both figures on the Rs 1,140 crore base, and establish the Rs 6.50 crore shortfall without any inside information at all. The funding arithmetic is the part that needs no access.

The outside reader listens for the vocabulary. A management team that talks about delivery against a fixed baseline, that quotes leverage with its basis named every time, and that will state a figure for what arrived rather than the word delivered, is a team running something like a real review internally. A team that answers with adjectives, that quotes a leverage number and moves on quickly, or that describes the purchase differently at each results presentation is saying something too. The review itself cannot be read from outside, but whether one exists can be. A company that runs them speaks in fixed baselines and named bases, and a company that does not speaks in adjectives.

A household version of the same instinct works on a much smaller scale. A shopkeeper who says the shop is doing well and, asked for more, can name the month it is being measured against, and by how much, and what it cost to get there, is running the same discipline on a kitchen table. Somebody who can only say it is going well may be doing just as well, but nobody, including them, can tell.

Harivansh Packaging Limited, Sundarban Polymers Private Limited, Devyani Kulkarni and Ashwin Rege are invented.
Educational material. Not advice on any investment, tax, budget or market position.