Deal Outcome or Process Quality: Which Can Be Judged

The outcome of a transaction is simply how it ended: it completed, it lapsed, it was terminated, or it was renegotiated and then closed. Process quality is whether the work was done properly, meaning conditions identified early, information controlled, the price basis stated, decisions recorded as taken. The two are independent, so a completion is not evidence of a good process, and a lapse is not evidence of a bad one.

What is a transaction outcome, and what states can it end in?

Start with a household rather than a company. A couple agree to buy a flat. The couple fix a price, pay a token amount, and wait while the bank looks at the paperwork. Six weeks later one of four things has happened. The purchase went through. Or the bank refused the loan, so the agreement expired on its own terms and nothing was bought. Or the seller discovered a title problem and used a clause in the agreement to walk away. Or a survey found damp in one wall, the two sides argued about it, the price came down, and the purchase then went through at the new number.

Four endings are the whole of what can happen, and none of them is a comment on how carefully the couple went about it. A transaction is the same picture with more zeros and more paper.

A transaction outcomeThe end state a transaction reaches. It is a fact about the transaction, not a verdict on the people who ran it. is a fact about the transaction and not about the people who ran it. The outcome is the answer to a single question: how did this end? There are exactly four answers, and each of them sits beside the process reading without saying anything about it.

The transaction completed. Every condition to completion was satisfied, the money moved, the shares changed hands, and the price was adjusted afterwards in the way the agreement said it would be. Completion is the ending the worked transaction below reaches.

The transaction lapsedA transaction that simply ran out. A condition was never satisfied by the date the agreement allowed for it, so the obligations fell away and nobody bought anything.. A condition was never satisfied by the date the agreement allowed for it, so the obligations fell away by themselves. Nobody had to do anything to end it. The lapse came about because something that had to happen did not.

The transaction was terminatedA transaction ended by somebody using a right the agreement gave them to walk away, rather than by a condition failing on its own.. Somebody used a right the agreement gave them to walk away. Termination is an act, where a lapse is an absence. Whether such a right exists at all, and what it costs to use, is settled by the paper and covered under transaction documentation. The ending in a termination is chosen by a party rather than produced by the calendar.

The transaction was renegotiatedTerms changed after signing, usually price or the shape of the payment, and the transaction then completed on the changed terms rather than the original ones. and then completed. Something turned up between signing and completion, the terms moved, and the transaction closed on the new ones. A reader who counts only completions will file this alongside a clean close, and a reader who counts only original terms will file it alongside a failure. A renegotiated completion is neither of those. Renegotiation followed by completion is a state of its own.

All four of those are outcomes, and only one of them is failure in the ordinary sense of the word. A transaction that lapsed because a condition identified in week one was never satisfied has behaved exactly the way it was designed to behave: the condition was there to stop the transaction if that thing did not happen, and it stopped it. Calling that a failure is like calling a smoke alarm a failure because it went off.

Notice also everything the outcome leaves out. The outcome carries no view on whether the price was sensible. The outcome carries no view on whether the two businesses fit. The outcome carries no view on whether the money could have been better spent elsewhere. The outcome is one word out of four, and it is the word most people reach for when they want a verdict.

What is process quality, and what is it actually made of?

The word quality invites a shrug. Most people, asked whether a piece of work was good, answer with a feeling: it felt organised, or it felt chaotic, or the team seemed on top of it. A feeling cannot be reviewed, cannot be argued with and cannot be improved, so a usable definition has to be built out of something firmer.

Think of a driving test instead. The examiner does not ask whether the drive felt confident. The examiner has a sheet with named items on it: mirror before signal, observation at the junction, position in the lane, control on the hill. Each item is either done or not done, and the sheet can be filled in by somebody sitting in the back who has never met the learner. The examiner's sheet is what a checkable standard looks like.

Process qualityWhether the work on a transaction was done properly, expressed as a short list of items that can each be answered by reading a document rather than by forming an impression. is the same idea applied to a transaction. Process quality is a short list of named items, each of which can be answered by reading a document that already exists. Every one of these items is built into the ordinary running of a transaction, so no new machinery is needed to read them, only a turn to look at work already done.

Were the audiences mapped from the contracts rather than from memory? Somebody had to write down every person and body that the transaction touches, and the honest way to build that list is to read the agreements the business already has and see whose consent or notice each one requires. The lazy way is to write down the audiences somebody remembers. Which of the two was done shows up as soon as the map is read beside the contracts.

Were the conditions identified before signing? A condition discovered in week eighteen was always there. The condition was simply not found. The test is not whether the conditions list was long but whether it was complete on the day of signing, and that is answerable by comparing the signed list with whatever turned up afterwards.

Was the price basis stated so that both sides meant the same number? A transaction can be agreed at a figure that each side understands differently, and the difference surfaces months later as an argument. The test is whether the wording fixes what the multiple was struck on, what it was struck against, and what is assumed about cash and debt.

Was the information controlled, and was the list of who had it kept current? Access to a live transaction is a thing that has to be granted, recorded and withdrawn. A list that was written once and never updated is not control. A log that shows who was added, when, and who came off it, is.

Were decisions and assumptions recorded separately? A decision logA dated record of what the transaction team decided, who decided it, and what was known at the time the decision was taken. records what was chosen, by whom, and on what was known at the time. An assumption registerA dated list of the things a transaction team is taking to be true without proof, each one marked with whether it can be checked and who is checking it. records what was being taken as true without proof. Mixing the two produces a document that reads like a story afterwards. A story is exactly what a reviewer must not be handed.

Every one of those five items is answerable today, by reading, without knowing anything at all about what happened afterwards. Answerability today is the property that makes process quality useful. Process quality does not wait for anything. The five answers were true in week three and they are true now, and they will still be true in five years when nobody remembers the transaction.

Why are the outcome and the process independent of each other?

Here is the sentence readers resist, put plainly before it is explained. Good work can be followed by a lapse, and careless work can be followed by a completion, and neither of those says very much about the other.

The resistance is natural. In most of ordinary life effort and result move together. Study harder and the marks improve. Cook carefully and the food is better. Transactions do not work like that, and the reason is not mysterious. The outcome depends on a set of things the process never claimed to control, and a process that is judged by them is being marked on somebody else's paper.

There are three of those things, and they show up on every transaction.

The first is an authority's view. A purchase may need an approval, and whether that approval arrives, and on what terms, is a decision taken by people the transaction team cannot instruct. A team can identify the approval early, prepare the material well and answer questions promptly, and every one of those is process. The decision itself is not.

The second is a counterparty's commercial position. On this transaction three conditions had to be satisfied, and two of them turned on consents from counterparties whose contracts were changing hands. Such a counterparty is entitled to say no, and it may say no for reasons that have nothing to do with either party to the transaction. Perhaps it is already unhappy with the pricing on that contract. Perhaps it has decided to bring the work in house. The team can find the consent early, assign somebody to own the relationship and go and ask in good time. The team cannot supply the answer.

The third is a state of the world. Between signing and completion the world carries on. A currency moves, a customer goes elsewhere, a plant is damaged, a market turns. Some of those events are large enough to trigger rights in the agreement and some are not, and none of them is caused by the quality of anybody's work.

The everyday version is an umbrella. Somebody looks at the sky, decides it might rain, and carries one. No rain falls. Was carrying the umbrella a mistake? Only if carrying it was meant to make rain. The decision was about what was known on leaving the house, and the weather was decided elsewhere.

Read the bottom row twice. Completion is mostly not in the team's hands. A measure the team cannot move is a poor measure of the team. The items in the left column are wholly in the team's hands, and a review can actually use them.

There is a softer version of the same point that is worth keeping. The two are independent, not opposite. Good process does raise the chance of a good outcome. Finding a condition early leaves nineteen weeks to satisfy it instead of two. Good process still does not decide the outcome, and a raised chance is not a decided result. The distinction between a raised chance and a decided result is the whole of the argument.

What do the four combinations look like when they are worked through?

If the two readings are independent, then both can be good, both can be poor, or one can be good while the other is poor. Four combinations follow, and all four occur. Listing them settles nothing, and each one has to be worked through before it means anything.

Good process, and the transaction completed. Everybody pictures this cell when they say a transaction was a success. The conditions were found early, the price basis was stated, the information was controlled, and the thing closed. The cell is also, oddly, the hardest to learn from. Nobody goes looking for problems in a story with a happy ending. A team in this cell should still be reviewed, and usually is not.

Good process, and the transaction lapsed. A transaction that should not have completed, and did not. The conditions were doing their job. Something that had to be true was not true, the agreement noticed, and the transaction stopped before anybody paid for a business they should not have bought. Read that way it is not a disappointment at all, it is an expensive test that produced the correct answer. Money and months were spent, and that cost is real. The alternative was buying a company on a false footing.

Poor process, and the transaction lapsed. The only one of the four that agrees with itself. Conditions were found late, the price basis was loose, the audiences were mapped from memory, and it fell over. Painful, but at least the lesson and the ending point in the same direction. Teams in this cell tend to correct.

Poor process, and the transaction completed. The fourth combination is the dangerous one. Everything was done badly and it closed anyway. The approval came through, the counterparties happened to say yes, and nothing went wrong in the world between signing and completion. Poor process with a completion teaches a team that the way it worked is the way to work, and that lesson quietly damages an organisation over years. The method gets repeated on the next transaction, a bigger one, and then on the one after that, bigger still. Nothing goes wrong until it does, and by then the method has been endorsed by three completions and is very hard to argue with.

The everyday version is somebody who overtakes on a blind bend and arrives safely. The driver will do it again. The safe arrival is the problem, not the reassurance.

Which of the two can be judged today?

Process quality, entirely. Every item on the list is answerable from documents that exist now, and the answers do not change when the ending arrives. The practical payoff of the whole distinction is worth saying twice: a review does not have to wait.

The outcome can be judged only once it has happened, and that sounds trivial until one counts how often people judge it before then. A transaction that is signed and waiting through its conditions periodThe stretch between signing and completion during which the conditions to completion have to be satisfied. Nothing has been bought yet and the transaction can still end. has no outcome yet. A signed transaction has an agreement and a set of conditions, and treating it as a completion because it looks likely to complete is the same error in a smaller form.

A third question sits behind both readings, and it is the one most readers actually came for. Was the transaction worth doing? No published figure settles that question, and no arithmetic anywhere closes the gap.

Consider what an answer would need. An answer would need to know what Harivansh Packaging Limited, an invented buyer, would otherwise have done with the money, a comparison against an alternative that never happened. An answer would need to know what the two businesses go on to make of each other over years. An answer would need a view on what the packaging market does next. None of those is in an annual report, and a number that is presented as if it settles the question is a number that has quietly assumed the answer.

The three rows above the total are all checkable, and the fourth is not, and the honest position is to be loud about which is which. Any account that called the transaction a good idea would be dressing up an opinion as a finding.

What happened on this transaction, read both ways?

Take one transaction and run both readings on it, keeping them strictly apart. Harivansh Packaging Limited agreed to buy the whole of Sundarban Polymers Private Limited, and the two readings answer two different questions about it.

The outcome reading

The transaction completed. The equity value paid was Rs 1,137 crore. The Rs 1,137 crore came from Rs 1,140 crore agreed at signing, adjusted upwards by Rs 12 crore and downwards by Rs 15 crore under the completion adjustments the agreement defined. And that is the whole of the outcome reading. Notice how little it says, and how much a reader wants to read into it.

The process reading, item by item

Now the other reading, using only what the earlier steps established, and answering each item from a document rather than from an impression.

Was the price basis stated? Yes. The enterprise value of Rs 1,320 crore was struck at 10.0 times Sundarban Polymers Private Limited's earnings before interest, tax, depreciation and amortisation (EBITDA) of Rs 132 crore, on a cash free and debt free basis. Because the basis was written that way, deducting that company's net debt of Rs 180 crore to reach the Rs 1,140 crore equity value was arithmetic that both sides had already agreed to, rather than a point to be argued at completion. 10.0 times Rs 132 crore is Rs 1,320 crore; less Rs 180 crore, that leaves Rs 1,140 crore. Anybody can check the arithmetic, and being checkable is the point.

Were the adjustments computable rather than arguable? Yes. The working capital pegA normalised level of working capital fixed in the agreement. The price moves by the difference between the peg and the actual level at completion, in either direction. of Rs 96 crore existed in the agreement before completion, so when working capital came in at Rs 108 crore the Rs 12 crore was a subtraction, not a negotiation. The same is true of net debt: Rs 180 crore assumed, Rs 195 crore actual, Rs 15 crore down.

Were the conditions identified before signing? Yes. Three of them were named at signing, and the nine week conditions period was spent satisfying conditions rather than discovering them. The difference between a conditions period that is administration and one that is a search sits exactly there.

Were the audiences mapped from the contracts? Yes. The two counterparty consents were on the list from the start rather than surfacing when somebody read an assignment clause in week seventeen.

Were the decisions and assumptions recorded apart? Yes, in two documents rather than one narrative.

Five answers, all yes, all read from paper. Set beside the outcome reading, they make two accounts of one transaction built from completely different evidence.

What if one consent had been refused?

A refused consent is where the distinction earns its keep.

None of them. Go back through the five and try to make one of them move.

The price basis was stated at signing, in writing, and a refusal in week twenty does not travel backwards and unstate it. The peg of Rs 96 crore was in the agreement, and it was computable whether or not it ever came to be computed. The three conditions were identified before signing, and the refusal of one of them is the proof that they were the right three rather than evidence that they were the wrong ones. The audiences were mapped from the contracts, and that is exactly how the counterparty came to be on the list to be asked. The decisions and the assumptions sat in two documents, and they still would.

The process was what it was, and the outcome would have been the other one. Same five answers, same five documents, same team, same weeks of work, and a different word at the end. If a review would have scored those two transactions differently, the review is not scoring the process at all. The review is scoring the ending and calling it the process.

One last thing before leaving the worked instance, and it is the name for the mistake. Judging a decision by how it turned out is called resultingJudging the quality of a decision by the quality of its result, rather than by what was known and considered at the time the decision was taken., and the term belongs to Annie Duke, in Thinking in Bets, 2018. Resulting is a useful word precisely because it gives a name to something people do without noticing, and a named habit is easier to catch than an unnamed one.

Both readings together still do not settle whether Harivansh Packaging Limited was right to buy Sundarban Polymers Private Limited. The two readings record what was paid and how the work was done, and what else that money could have bought is in neither of them.

How should a review keep the two apart?

A reviewA structured look back at how a transaction was run, done to improve the next one rather than to allocate credit or blame for this one. is done to improve the next transaction, not to allocate credit for this one, and it holds that purpose only if it is run in a fixed order. Three steps, and the order carries almost all of the value.

Step one. Read the decision log and ask, of each decision, whether it was reasonable on what was known at the time. Not whether it worked. Whether it was reasonable then. The log carries dates for exactly this reason: a decision taken in week four is judged against week four, and information that arrived in week eleven is not admitted.

Step two. Read the assumption register and ask which assumptions proved false, then ask which of those could have been checked at the time. Step two is the step that finds real improvements. An assumption that proved false and could not have been checked is bad luck. An assumption that proved false and could have been checked in an afternoon is a finding, and it is a finding whether the transaction completed or not.

Step three, and only now. Record what happened. The outcome goes into the review as its own separate finding, written after the other two, and it is not allowed to travel back into them.

A reviewer who knows the ending reads everything as leading to it, so doing the third step first contaminates the other two. Contamination is not a failure of honesty and it is not fixed by trying harder. Reading simply works that way. Once the lapse is known, the week four decision to spend two days on a counterparty relationship reads as time wasted; had it completed, the identical decision reads as foresight. Same paragraph, same two days, two readings, decided entirely by a fact the reviewer picked up before opening the file.

The everyday version is marking an exam. An examiner who is told the student's final grade before marking the first paper will mark differently, and serious examinations are marked blind for that reason. A transaction review that starts from the ending is a marked paper with the grade written on the front.

What happens to a team judged on outcome alone?

Measure a team on how many transactions complete and completions rise. Two other things arrive with them, and the second is the expensive one.

The first is that selection changes. A team that is scored on completions stops taking transactions that might lapse, and transactions that might lapse are not the same set as transactions that should not be done. The most valuable purchase available may be the one with a difficult consent attached to it. A team protecting its record will quietly prefer the smaller, simpler, safer transaction that nobody will criticise, and the organisation ends up with a portfolio chosen for its likelihood of closing rather than for what it is worth.

The everyday version is a student who picks the easy elective to protect a grade average. Nothing dishonest has happened, and something has been lost.

The second effect is that the record goes quiet, and that effect costs real money: judging by outcome destroys the record that would have let anybody improve the process. If a decision log is going to be read afterwards as a list of errors, people stop writing candid decision logs. If an assumption register is going to be produced as evidence that the team was reckless, the register becomes a list of assumptions nobody minds being seen holding. The residue is a set of documents that is technically maintained and carries no information.

Notice how that closes the loop. The whole reason process quality is checkable is that these documents exist and say what actually happened inside the team. Judge on outcome and the documents empty out, and once they are empty, process quality becomes unmeasurable, and the only thing left to measure is the outcome. The organisation has argued itself into a corner where the one thing it can see is the one thing the team does not control.

The instinct behind outcome measurement is worth being fair to. Outcomes are cheap to observe, they arrive with a date attached, and nobody argues about which one happened. Process quality takes a reader, some hours, and a willingness to be told something inconvenient. The convenient measure wins by default unless somebody insists otherwise, and that insisting is what a review discipline is.

How does a transaction team actually use this split?

The distinction earns its place only if somebody does something with it, so here is what that looks like in practice for four different readers.

The chief financial officer. Devyani Kulkarni, who holds that post at Harivansh Packaging Limited, has to answer a board that will ask whether the transaction went well. The useful answer separates the two. The transaction completed at Rs 1,137 crore, and that is the outcome. The five process items were all answered yes, from documents, and that is the quality reading. And whether the purchase was worth making will be visible only in what the combined business does over years, a third question and not one a completion answers.

The person who leads the transaction team. Ashwin Rege, in that role here, uses the split to decide what to bring to a review and what to leave out. He brings the decision log and the assumption register, in that order, and he brings the ending last, on its own sheet. He also uses it forwards rather than backwards: if the five items are what a review will read, then the five items are what the team maintains from week one, and maintaining them is not extra work done for a reviewer but the work itself.

A lender. A bank lending into a purchase is exposed to a borrower's future transactions, not just this one. A borrower whose transactions all complete tells the lender very little. A borrower that can produce a decision log and an assumption register, and can show a transaction it walked away from and why, is describing a repeatable process. The second borrower is easier to underwrite, and the reason is that a process can be relied on where a run of endings cannot.

An analyst or an investor reading from outside. Very little of the process reading is visible from outside a company. Visible from outside are the outcome and whatever the company chose to disclose. The outside reader's version is therefore a discipline of restraint: a completion is not evidence of capability, a lapsed transaction is not evidence of incompetence, and what repays attention is which companies talk about how they run transactions rather than only about which ones closed.

The bottom row is the uncomfortable one, and it explains why outcome talk dominates public discussion of transactions. The outside world can see endings and cannot see logs, so it discusses endings. The silence is a limitation of the vantage point, not evidence that endings are what matter.

What do the outcome and process quality have in common?

More than a comparison usually admits. Both are descriptions of the same transaction, built from the same weeks of the same people's work. Both are recorded, and both belong in a review. Both are facts rather than opinions once their meaning has been fixed, and the process reading is a five item list rather than a feeling for exactly that reason.

And both of them stop short of the same line. The outcome does not say whether the purchase was wise. The process reading does not say either. A team can run an exemplary process on a transaction that should never have been contemplated, and the five answers will all be yes.

The two readings answer two different questions, and the value of keeping them apart is that each one then answers its own question cleanly instead of both of them answering neither. One of four words answers what happened. Whether the work was done properly has five answers, read from paper. Whether it was worth doing has no answer yet, and possibly never will.