The Executive Summary: Saying It Before You Explain It

An executive summary states the answer, the two or three supports, and the one condition that would change it. A summary runs to a single printed side or less, and it is written last, once the answer is settled. The summary is not an introduction and it never sets up what follows: a reader who reads only the summary must end up with a true picture rather than a partial one.

Here is what sits underneath that. Most readers of most documents read only this part. Reading only the top is not a failure of attention, and it is not laziness. The habit is a sensible allocation by somebody who has eleven documents and one afternoon, and who has learned that eight of the eleven will not change anything they do. Writing the summary as though it will be the only part read is simply writing it for what actually happens.

One assignment runs through this guide. The Kavery research desk, inside Kavery Capital Services Private Limited, produced a 1,400 word memo about Meenakshi Tubes Private Limited, a borrower the desk already lends to. Sharada Iyer wrote it, Prakash Nadar reviewed it, and Latha Menon took the decision at the end. On top of that memo sat 180 words. The 180 words on top are the subject; the memo underneath is not.

The product the borrower manufactures does not matter anywhere below, and the structure of the memo underneath is set out under the finance memo. The subject here is a form rather than a business. Substitute a bond, a fund, a warehouse or a single machine for Meenakshi Tubes Private Limited and not one rule moves. A summary has four parts, one length, three things that must be kept out and one test that decides whether it works, and a summary that reads beautifully can still fail every one of them.

What is an executive summary for, and who actually reads it?

An executive summaryA short standalone statement of the answer, its supports and its condition, placed at the top of a longer document. exists because attention is scarce and unevenly distributed. An executive summary is not a courtesy to a busy person. The summary is the part of the document that does the work when nobody reads the rest, and nobody reads the rest most of the time.

Picture something smaller than a committee. A hospital sends somebody home with a discharge note. The full file runs to forty printed sides of charts, readings and observations. On the front there is one sheet: what was wrong, what was done, what to watch for, and when to come back. Everybody who matters after that point reads the one sheet. The forty sides exist so that the one sheet can be trusted and challenged, and almost nobody opens them. The summary is not a shorter version of the document; it is the part of the document that will actually be used.

Once that is accepted, three things follow immediately and none of them is optional. The summary has to be complete on its own terms. There is nothing beyond it for the reader who stops. There is no later paragraph where the caveats live, so the summary has to be honest about what it does not know. And patience runs out earlier than writers believe, so the finding has to arrive before the reader runs out of it.

Who reads it, in practice? Somebody deciding whether to spend more attention on the file. Somebody who has to repeat the finding to a third person in one sentence tomorrow. Somebody skimming to check whether their own position has changed. All three of them are doing the same thing: taking the top of the document as the document. What they do immediately afterwards is act, delegate, or move on, and all three of those happen without the rest of the file being opened.

How to Write an Executive Summary for Financial Analysis: What Are the Four Things It Carries?

Four parts, and nothing else. The answer is the finding stated as a finding. The supports, two or three of them, are each a reason the answer holds. The conditionThe one thing that, if it turned out differently, would change the answer rather than merely soften it. is the one thing that would change the answer. And the action is what the document proposes, in one line, after which the summary stops.

A finding stated as a finding is exposed, so the answer is the hardest of the four to write. Operating margin is 10.8 per cent, below the 11 per cent internal floor and down from 12.0 per cent a year earlier. A finding stated that plainly can be wrong in a way that the margin was examined never can. Writers drift towards the second kind without noticing, and the drift is what turns a summary into something nobody can act on.

The supports are the reasons, not the workings. Two is usually right and three is the ceiling. A fourth support does not strengthen an answer; it signals that the writer could not decide which two mattered, and a reader reads that signal accurately. Supports are chosen by asking which reasons a challenger would attack first, not by listing what took the longest to produce.

The condition is the part that gets dropped. The condition is the sentence that says what would have to be different for the answer to change, and it is dropped for an understandable reason: it weakens the paragraph just written. Leaving it out does not merely shorten the summary. Dropping the condition converts a qualified finding into an unqualified one, and the reader has no way of telling from the summary that anything was removed.

The action is one line. The proposal, and then a full stop. A summary that argues for the action has become a second memo; a summary that omits it leaves the reader holding a position with nothing to do about it.

How long should it be, and what does the limit actually enforce?

One printed side or less. On the case, 180 words sat above a 1,400 word memo, a compressionThe ratio between the length of the summary and the length of the document sitting under it. of nearly eight to one. Eight to one is not a rule of nature and nobody will check it, but it is a useful sanity mark: divide the document by eight and the result is roughly the space available.

The limit is not about brevity, and this is the part people get wrong. A busy reader is not saved much by two hundred fewer words. A hard limit is valuable because it forces a choice, and the choosing is the entire work the summary exists to do. Give a writer two printed sides and they will include everything defensible. Including everything defensible is precisely what produces a summary nobody can use. Give them one printed side and they have to decide which two supports matter. Deciding which two supports matter is a real analytical judgement, and nobody else in the process is making it.

One thing the limit does not do is shrink the document proportionally. Cut every part of that 1,400 word memo to one eighth of itself and the result is a three word answer, a four word statement of what is proposed, and eighty words of supporting case: a damaged memo rather than a summary. The summary is a different selection from the document, not a smaller copy of it. The figure below sets the two allocations against each other, and the two rows that fall to zero are the ones worth staring at.

What must never go into an executive summary?

Three things, and they arrive in the same order every time. Background. Method. Anything unresolved.

Background is the easiest to spot and the easiest to write, and that is why it wins. Background feels like courtesy: remind the reader who the borrower is, when the file was opened, why the question was asked. Every one of those sentences is a sentence the finding did not get. The reader either already knows the context or can get it from the first paragraph below the line, and in neither case do they need it before they are told what was found.

Method is the subtler one, and the one experienced writers lose to. Method is how the work was done: the sample, the source ranking, the way a clash between two documents was settled, the checks that were run. Method is the most defensible material in the whole file and it belongs entirely below the line. Method is what a reader turns to when they have decided to challenge the finding, and a summary is written for the reader who has not started challenging it yet.

Anything unresolved is the third, and it hides behind good intentions. A sentence saying that two items in the source log are being re-checked, or that a figure is awaiting confirmation, feels scrupulous. In a summary an open item is corrosive. A reader can neither act on it nor ignore it. Either the open item changes the answer, in which case it is the condition and gets written as one, or it does not, in which case it belongs below the line with everything else that did not change the answer.

The last row of the table looks like a contradiction and is not, so it is worth slowing down on. Saying how the desk arrived at a sixty forty split between fixed and variable cost is method, and it stays out. Saying that the split was chosen rather than filed is a qualifier on the answer itself, and it stays in. The test is not whether a sentence is about the work; it is whether removing it changes how solid the answer looks. On the case that qualifier cost fourteen words and it earned every one of them.

Why is it written last, and what happens to one written first?

Written last, always, and this is not a scheduling preference. A summary is a selection from a finished argument. Until the argument is finished there is nothing to select from, so a summary written first is not a summary at all; it is a hypothesis wearing the clothes of a conclusion.

Think about a household deciding whether to change schools. If somebody writes the note on the fridge in week one saying the decision is to move, everything gathered in weeks two and three gets read against that note. Evidence that fits is confirming; evidence that does not is an exception to be explained. Nobody is being dishonest. The note simply became the thing the enquiry had to agree with, and by week four it had stopped being a draft and started being the answer.

A summary written first does not stay a harmless placeholder; it quietly becomes the standard the rest of the analysis is measured against. The drift is a real failure mode and it is invisible from inside. At the end the document is perfectly consistent. Everything agrees with everything else. Nobody can see that the agreement was arranged in week one.

There is a legitimate cousin of this that is worth distinguishing. Writing a one line working hypothesis at the start, clearly labelled as the proposition the work is trying to disprove, is good practice. The difference is not the timing, it is the status. A hypothesis invites its own destruction. A summary written first invites confirmation, and it gets it.

How is it different from an introduction or an abstract?

Three documents, three jobs, and confusing them is the commonest fault of all. An introductionA passage that sets up what is coming next, which a summary must never do. sets up what follows. An abstractA description of what a document contains, rather than a statement of what it found. describes what the document contains. A summary states what the document found. Only the third of the three can be read on its own without leaving the reader worse informed than if they had read nothing.

Here is the diagnostic, and it takes four seconds. Read the opening sentence and ask what the reader now knows that they did not know before. If the answer is what is coming next, it is an introduction. If the answer is what the document contains, it is an abstract. If the answer is a finding they could repeat to somebody else tomorrow, it is a summary.

The abstract is the dangerous one because it is the trained default. Academic writing rewards abstracts, and most people learn to write summaries by writing abstracts and shortening them. An abstract can be entirely accurate, cover every section faithfully, and still leave a reader who reads only it holding nothing they can use. The defect is not a small one. It is the whole failure, dressed as competence.

How can a writer tell whether a summary works?

One test decides it. Take the summary, remove everything underneath it, and read what is left as though it were the whole document. Ask whether the picture the reader now holds is true. Asking that question is the standalone testStripping the rest of the document away and asking whether a reader left with only the summary is misled., and it is the only test a summary has to pass.

Notice what the standalone test does not ask. The test is not asking whether everything in the summary is true; a failing summary is usually true throughout. It is not asking whether everything important is present; nothing 180 words long contains everything important. It asks one thing only: does what is absent mislead? A fact left out that would not have changed the reader's picture is a fact correctly left out. A fact left out that would have changed it is a failure, however tight the word count was.

Two kinds of absence fail the test, and they fail differently. The first is a load-bearing input left unnamed. On the case, the split between fixed and variable cost was chosen by the desk rather than filed, and the 5.0 per cent scenario is built on it. A summary that gives the scenario and never says the split was chosen presents a constructed number as a measured one. The second is order. Order is the more common of the two failures by a wide margin, and the two versions set side by side below show what it costs a reader.

What did 180 words above a 1,400 word memo actually carry?

The length is worth predicting first. The allocation below shows what the four parts did with the space.

The summary on the Kavery memo ran to 180 words. The summary carried the four parts and one qualifier. The allocation is itself a judgement somebody made, so the arithmetic of the space is worth seeing laid out.

Look at what is not there. No source log, though the memo carried fourteen items. No review findings, though Prakash Nadar produced them. No alternatives considered. Not one word about how the work was done. The summary spends more than a quarter of its words on the condition and less on the answer itself, and that ratio is the mark of a summary written by somebody who expected to be read only once.

Apply the standalone test to it. A reader who reads only those 180 words knows the margin, knows it is under the floor, knows the two reasons, knows the one thing that would change the answer, knows that a key input was chosen rather than found, and knows what is being proposed. Nothing they did not read would change any of that. It passes.

Do the two versions really cover the same ground?

The two versions do, and the pair below is built so that the claim can be checked rather than taken on trust. Both are 180 words. Both carry the same five parts. Each part carries the same number of words in both, so the two texts are the same content in the same quantities, and the sequence is the only thing that differs. Answer-first ordering is not a matter of taste here: Barbara Minto set it out as the Pyramid Principle in 1978, and the argument for it is exactly what the pair below demonstrates.

The dashed line in each panel sits at word 129 of 180, a real sentence boundary in both versions. The comparison that matters is between what a reader holds after stopping at the line in one version and what a reader holds after stopping at the line in the other.

The two closing sentences of version B are word for word the two opening sentences of version A. The overlap is not a trick of the illustration; it is what happens when a writer builds the argument and then narrates the build. The finding and the proposal are in the last 51 words of one version and the first 51 words of the other, and everything else is identical.

Where does a summary shorten, and where must it not?

Shortening is the whole job, so it helps to know which cuts are free and which are not. Three cuts are free. Cut adjectives, cut connective throat-clearing, and cut precision nobody will act on. Three cuts are never free. Never cut a condition, never cut a qualifier that changes how solid a number looks, and never cut the direction of a change.

Rounding sits on the line between the two lists and deserves a rule of its own. Rounding Rs 2,96,80,000 to about Rs 3,00,00,000 saves two words and usually costs nothing. Rounding costs something the moment a threshold sits inside the gap that was rounded away, and that is exactly what happened elsewhere in this case: 10.8 per cent and 12.5 per cent sit on opposite sides of an 11 per cent floor, and no rounding of either is innocent. Round the number when nothing anybody does turns on the difference, and never round the qualifier at all.

Plain language is a shortening that costs nothing and is almost always available. On the case, the non-technical version of the answer reads: for every Rs 100/- of sales, Rs 10.80 is left after running costs, against Rs 12.00 a year earlier. The plain sentence carries exactly the same content as an operating margin of 10.8 per cent against 12.0 per cent, it is no longer, and it reaches a reader who does not work with margins. Nothing was surrendered to get it.

One honest tension is worth naming, and version A has it. Its qualifier and pointer sit in the last 31 words, below the line at which many readers stop. A writer who thinks the chosen cost split is truly load-bearing should move that qualifier up beside the condition and push the pointer to the very end. There is no rule that settles this. The ordering question inside a summary is decided by asking which absence would mislead a reader who stops, and then putting that thing above where they stop.

How does a lender, an analyst or somebody working alone use the top of a document?

A credit officer with eleven files on a Tuesday reads eleven summaries and opens two of them. The summary is not a preview of the file for that person; it is the triage instrument that decides whether the file gets opened at all. Which means a summary that fails to state a finding does not get a gentle second chance. A summary with no finding gets sorted into the pile of things that did not seem to say anything, and the work underneath it never surfaces.

An analyst covering a sector uses the top of a document differently: as a thing to disagree with. They read the answer, form a reaction, and go straight to the condition to see whether the writer identified the same weak point they did. If the condition is missing, the analyst concludes the writer did not look, a harsher judgement than being wrong about it. A stated condition is what tells a sceptical reader that somebody thought about the downside before they did.

Somebody working alone, with no committee and no reviewer, gets the most from this of anybody, and it is worth spelling out because it is easy to assume the form is corporate furniture. Writing the summary last forces a one sentence statement of what the work concluded. Plenty of analyses turn out not to have a conclusion, and that becomes apparent at exactly the moment the first sentence has to be written. The summary is a check on the work as much as a wrapper for it.

One practical habit closes the loop. After the summary is drafted, it goes to somebody who has not read the file, who is asked to say back what it said. If what comes back is a list of what was looked at, the draft is version B. If what comes back is a finding with a caveat, the draft is version A. The exercise takes four minutes and it catches the one failure that rereading one's own words never will.

The Kavery research desk, Kavery Capital Services Private Limited, Meenakshi Tubes Private Limited, Sharada Iyer, Prakash Nadar and Latha Menon are invented.
Educational material. Not advice on any investment, tax, budget or market position.