How to Build a Transaction Document Map, Step by Step

A transaction document map is one sheet that lists every document in a transaction, puts a clause reference beside every figure, and records the drafting choices that decide what money actually moves. The map is built in six steps. On this transaction it places twenty six entries, five defined terms and two decisive choices, and it answers any question about the paper in under a minute.

Three sentences are the whole of it, and everything below is how the map is built. A summary is written to be read once. A map is built to be looked up a hundred times, so a map is neither a summary nor a memo. The difference sounds small until somebody asks, in a meeting, what was actually paid, and the answer either arrives in eleven seconds or arrives after two hundred printed sides have been opened and scrolled.

Why build a map at all, when the document is right there?

Think about the folder of paper that comes home with a household after it buys a flat. There is the sale deed, the loan sanction letter, the schedule of the property, the receipt for the stamp duty, a letter from the society, and a sheet of undertakings somebody signed at the sub registrar office. Every one of those is a document. The price is written in one of them. The consequence of the seller leaving maintenance dues unpaid is written in another. The date by which the loan has to be drawn is in a third. And when a question comes up two years later, the household does not read all six documents again. Somebody sensible wrote a single sheet, at the time, listing what is in which document and where.

A transaction map is that same sheet, built for a transaction rather than a flat, and built by method rather than by luck. The map contains nothing the documents do not. Its whole value is speed. Every figure on it becomes findable in seconds instead of minutes.

There is a second reason, and it matters more. Reading a document end to end gives the order the drafter chose. The drafter's order is a legal order, not a commercial one. The amount paid, the amount recoverable and the amount funded can sit ninety printed sides apart, in three different documents, under headings that do not name them. A map re-sorts the same content into the order a commercial reader needs. The commercial order runs by the job each figure performs rather than by the place the drafter put it.

What does the map assume is already known?

The map assembles rather than explains. Every mechanism the map records is covered separately, so when a step points at something, it points and moves on. A step that starts explaining what a threshold does has quietly stopped being a step and become a lesson, and the map gets longer while getting less useful.

Three things are taken as settled. First, that when a document defines a word, that definition is the only meaning the word carries inside it, however everyday the word looks from outside. Second, that this transaction has a complete set of documented figures, running from an enterprise valueThe whole business measured before its borrowings are taken out. It is not the amount handed to the sellers, which is what remains after the borrowings come off. of Rs 1,320 crore at one end to a Rs 113.70 crore amount held in cash at the other. Third, that the map records where things are and what they are worth, and forms no view at all about whether any of it was well drafted.

What are the six steps, in order?

1. List every document, not just the main oneThe agreement, the schedule of exceptions to the promises, the escrow arrangement holding money in cash, the restriction on the seller, any letter signed alongside, and every schedule and attachment at the back.
   Leaves behind: one list. Nothing on it has been read yet.
2. Record the four clause groups in each documentWhat is being sold, the price and how it is arrived at, the promises, and the limits on the promises. The step records locations, not opinions.
   Leaves behind: a location for every kind of clause needed later.
3. Trace every figure to the clause that defines itWrite the clause reference beside the number, in the same row, every time. A number with an empty box beside it is the whole reason for doing this.
   Leaves behind: a clause reference beside every figure, or a visible blank.
4. Build the definitions registerList every defined term that appears inside a clause that moves money, and write down where the document defines it.
   Leaves behind: five terms on this transaction, each with its home clause.
5. Build the conventions registerRecord every place where the document picked one of two ordinary ways of doing something, and record which one it picked.
   Leaves behind: two lines on this transaction, and both of them decide money.
6. Record what could not be found, by nameA clause looked for and not located. A term used but never defined. A figure with nothing behind it.
   Leaves behind: five named gaps, written on the map rather than left off it.

Step one, in a little more detail

The instinct is to open the definitive agreement and start. Resist it for twenty minutes and write the list first. On this transaction the list runs to six kinds of document: the agreement itself, the schedule of exceptions the seller attaches to its promises, the arrangement under which cash is held, the restriction that stops the seller rebuilding the same business, any side letterA short separate document, signed at the same time as the main one, that changes or adds to something inside it. signed at the same time, and every schedule and annexureA block of detail attached at the back of a document, carrying lists, calculations and formats that would clog the clauses if written inline. attached at the back.

A figure quoted in one document is very often defined in a different one, so a map that is missing a document cannot trace those figures at all, and gives no sign that it cannot. The absence of any signal is what makes the omission expensive. A map with five of six documents on it looks exactly like a finished map. The incomplete map has rows, it has references, it looks complete, and the figures it silently cannot check are the ones sitting in the document nobody listed.

The listing is done mechanically. The signing bundle, the index and the signature sheets between them hold every separate thing that somebody signed or attached, and each of those things is written down. Reading starts then, and only then.

Step two, in a little more detail

Every transaction document, whatever it is called, answers four questions in some order: what is being sold, what is being paid and how that amount is arrived at, which promises the seller gives, and what limits sit on those promises. Step two is nothing more than writing down where each of those four sits in each document on the list.

Step two is pure locating and no analysis happens inside it. A step with no analysis in it feels skippable, and step two is the one people skip and then regret. The price clause is not being read to see whether the price is fair. The record simply says that the price clause sits at a particular place in a particular document and that the mechanic adjusting it sits somewhere else entirely. Half an hour of that saves every subsequent scroll.

The four groups are not equally sized and they are never in the same order twice. On this transaction the limits sit in one document and the cash that funds them sits in another. A location record catches that kind of split. A linear read does not.

Step three, in a little more detail

Now the map starts earning its keep. Take every number that anybody has quoted about this transaction, from the headline down to the smallest threshold, and put it in a row with the clause that defines it beside it. Not the clause that mentions it. The clause that defines it.

The distinction between defining and mentioning is the whole step. A number can appear in six places in a bundle and be defined in one. The other five are references, and a trace to a reference is a trace to nothing. A reference moves when the definition moves, and nothing announces it.

A figure with no clause beside it is a figure somebody said out loud, and the entire value of the map comes from having no such entries left on it by the time the work stops. Numbers travel through a transaction by conversation. Somebody says a figure in a call, it gets into a note, the note gets into a deck, the deck gets quoted in a board paper, and by the time it reaches the analyst it has the confidence of a documented figure and none of the backing. The clause reference column is what separates the two.

Step four, in a little more detail

Here is the step almost everybody skips. Go back through every clause that moves money, and list the defined terms sitting inside it. Then write down where each of those terms is defined. Not what it means. Where it is defined.

A clause that moves money is only as reliable as the defined terms inside it. Step four catches more errors than the other five put together. The reason is simple and slightly unfair. A defined term looks like an ordinary word. Cash looks like cash. Working capital looks like working capital. A reader glides over both without noticing that the document has given each of them a specific meaning, and that meaning is where the money is.

On this transaction, five defined terms sit inside clauses that move money: working capital, net debt, cash, earnings before interest, tax, depreciation and amortisation (EBITDA), and loss. The meaning each of those terms carries in this document is settled under the definitions that govern them, and is not rebuilt in the map. The map records the location of each term and the amount that turns on it.

Step five, in a little more detail

A drafting convention is a place where the document could reasonably have done a thing in one of two ordinary ways, and picked one. Both ways are perfectly normal. Neither is a mistake. And the pick changes the answer.

Step five is to hunt for those places and write down, for each one, what the two ways were and which one the document took. That is all. The register does not evaluate the choice, does not say whether it was the right one, and does not say whether it would hold up. The conventions register is always short, it is where the largest surprises on any transaction live, and it is the only part of the map that predicts an outcome rather than describing an arrangement.

On this transaction there are two entries. The first is whether the threshold that counted claims must clear is a level that, once passed, opens up the whole claim, or a slice that is simply deducted from it. The second is which of the seller promises the limits actually reach. A promise the limits do not reach behaves completely differently from one they do. Each of those two mechanisms is covered separately, and the map records the choice and the money, not the mechanism.

Step six, in a little more detail

The last step is the one that feels wrong to do and is the one that separates a finished map from an abandoned one. The missing items are written down, by name.

Three kinds of entry go here. A clause looked for and not located. A term the document uses but never defines. A figure that arrived with nothing behind it. Each gets a line, each gets a name, and none of them gets a guess.

An empty line on a map is information, and a map with no gaps recorded on it is a map somebody stopped filling in rather than one somebody finished. A recorded gap is not a rhetorical flourish. A reader picking up a map with a gaps register knows exactly which parts of the paper it covers and which it leaves out. A reader picking up a map with no gaps register has no way to tell the difference between a transaction with nothing missing and a mapper who gave up at four in the afternoon.

The gaps also route the next conversation. Every line on the register is a question for somebody. On a live transaction the gaps register is the most useful list on the whole sheet, and the only one that generates work.

What does a finished map look like, group by group?

Harivansh Packaging Limited, an invented buyer, has bought all of Sundarban Polymers Private Limited, an invented target, and every figure below was built for teaching. The map itself follows, group by group, and it reads as running text without looking at any drawing at all.

The price group

Sundarban Polymers earned EBITDA of Rs 132 crore. The transaction applied a multiple of 10.0 times to it, giving an enterprise value of Rs 1,320 crore. Take off the Rs 180 crore of net debt that Sundarban Polymers carried. The equity value, the amount that actually reaches the sellers, is Rs 1,140 crore. Enterprise value is not the amount paid to sellers, and a map that lets those two share a column without labelling them has already produced the mistake people make most often with deal figures. Five entries, all in the price clause group.

The completion mechanic group

Two adjustments run against the Rs 1,140 crore, and the map keeps them in a table because they have rows.

Six entries land in this group and the adjusted equity value at the foot of it is Rs 1,137 crore. The reason both halves are computed separately, and the working of the mechanic itself, belongs to completion accounts rather than to any map.

The contingent group

A further Rs 60 crore is payable as an earn-outPart of the purchase price held back until a stated result arrives. If it never arrives, the money is never owed. and the condition attached to it is EBITDA of Rs 145 crore in the twelve months following completion. Set that condition against the Rs 132 crore Sundarban Polymers actually earned and it sits 9.8 per cent higher. Should it be met, maximum equity value becomes Rs 1,197 crore and maximum enterprise value Rs 1,392 crore. Five entries.

The limits group

All four figures in this group are cut from one base, the adjusted equity value of Rs 1,137 crore, and the map names that base in every row. Identical percentages taken against the Rs 1,140 crore headline land on different rupees. The de minimisA floor applied to each claim on its own. Anything beneath it drops out before the totals are even added up. at 0.1 per cent is Rs 1.14 crore. The basketA running total of counted claims that must be exceeded before any payment falls due. A genuine, provable loss beneath it recovers nothing. at 1.0 per cent is Rs 11.37 crore. The amount held in escrowMoney set aside with a third party at completion, so that if a claim succeeds later there is cash already sitting there to pay it from. at 10.0 per cent is Rs 113.70 crore, held for eighteen months. The cap at 20.0 per cent is Rs 227.40 crore. Five entries.

The protection group

Three entries, and all three are constructed for this sequence rather than taken from any market. A break fee of Rs 22.74 crore, being one fiftieth of the adjusted equity value. A reverse break fee of Rs 45.48 crore, being exactly double it. And an exclusivityA window during which the seller agrees to talk to nobody else. It buys the buyer time to spend money on checking. window of thirteen weeks. The map carries the constructed label on all three, in the same row. A reader who lifts the figure without the label has taken an invented number for a real one.

The calendar

Twenty two weeks ran from term sheet to completion on this transaction, of which the conditions period was nine. Two entries, and both of them are this transaction's own elapsed weeks rather than any statement about how long a transaction takes.

The three registers

Beneath the six groups sit the three registers, and this is where the map stops describing and starts predicting.

The definitions register carries five terms: working capital, net debt, cash, EBITDA and loss. Working capital sits inside the peg clause and moves Rs 12 crore. Net debt sits inside the completion adjustment and moves Rs 15 crore. Cash sits inside net debt, so it moves that same Rs 15 crore rather than a separate amount. EBITDA sits inside the earn-out condition and moves Rs 60 crore. And loss sits inside every one of the four limit figures, so it governs the whole Rs 227.40 crore ceiling. Three of those terms move Rs 87 crore of price directly between them, and the fifth decides what the ceiling above them applies to at all.

The conventions register carries two lines and both of them decide money. The first is whether the Rs 11.37 crore threshold, once cleared, opens the whole claim or is simply deducted from it: on a Rs 15 crore claim that is the difference between Rs 15 crore and Rs 3.63 crore. The record rounds the gap between the two to 76 per cent of the claim; computed on the figures it is 75.80 per cent. The map prefers the rupee amount, Rs 11.37 crore, and a rupee gap does not change when somebody picks a different claim size to describe it with. The second line is which of the seller promises the limits reach. On a Rs 8 crore loss that line is the difference between recovering Rs 8 crore and recovering nothing.

Add those two lines together and Rs 19.37 crore turns on drafting across the two claims. Set that beside the completion mechanic. The mechanic moved the price by Rs 3 crore net, so the two lines are worth more than six times the whole adjustment machinery. The two lines are not worth more than the Rs 1,320 crore in the price group, and any map that claims so is overselling itself. Their real claim is narrower: everything still capable of moving after the parties have signed.

The gaps register carries five lines, and the first of them is about the multiple. Rs 1,392 crore of maximum enterprise value reads as 10.55 times the Rs 132 crore Sundarban Polymers actually earned, and as 9.60 times the Rs 145 crore the earn-out is conditioned on. Both readings are arithmetically correct and they describe different things, so no note may quote a multiple for this transaction without saying which EBITDA it used. The other four gaps are recorded below.

What does the finished map make possible?

The map is a working document, not a deliverable. Nobody signs it, nobody sends it to the other side, and its quality is measured in one thing only: how fast it answers a question. A map that takes four minutes to answer a question about the paper is not a good map that is slightly slow, it is a summary that has been formatted to look like a map.

Four questions in particular. A headline never answers any of them, and a finished map answers all four in under a minute.

Two things about that table are worth saying out loud. Both are places a reader can go wrong at speed. The first is that Rs 113.70 crore is exactly half of Rs 227.40 crore, and the arithmetic tempts a reader into saying that half the ceiling is cash. The escrow funds the first Rs 113.70 crore of any recovery rather than half of each one, and the difference between those two readings is covered under limitation. The map records the amount and points there.

The second is that the earn-out sits outside all four answers. Rs 60 crore is payable only if a stated result happens, so what was actually paid has one answer today and possibly a different one in a year. The map carries both, in the contingent group, with the condition beside them.

The five lines are worth walking through. Each one shows what a gap looks like when it is recorded honestly. The first is the multiple, already dealt with. The second and third are the two conventions: the map records what each choice would produce, in rupees, and records that it does not know which was taken. The fourth is a date: the eighteen month period during which cash sits set aside is recorded, and the outside date for bringing a claim is not, so the map says so rather than assuming they are the same period. The fifth is an accounting figure. The Rs 1,140 crore paid against net worth of Rs 320 crore leaves Rs 820 crore before any allocation to identified intangibles, and that Rs 820 crore was struck on the headline rather than on the Rs 1,137 crore actually paid. The map records the base it was struck on, notes that it was not restated, and routes the allocation itself to the accounting layer where it belongs.

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