The Pyramid gives you structure; the "so what?" discipline gives you content that earns its place. It is the single habit that most separates an analyst's writing from a data dump, and it is brutally simple: after every sentence, ask "so what?" If the sentence does not already answer it, either make it answer or cut it. The discipline attacks two failures at once: sentences that state facts without implications, and sentences that exist only to clear the throat.
What is the difference between an observation and an insight?
Start with the first, because it is the deeper one. There is a categorical difference between an observation and an insight, and beginners fill memos with the former believing they are delivering the latter. An observation is a true statement about the data. An insight is what the observation means for the decision. "Revenue grew 14% this year" is an observation; a reader's honest response is so what: is that good, why did it happen, what does it tell me? The insight is the answer to that: "Revenue grew 14%, and critically it was all volume, with flat realizations — the company is winning units without cutting price, which is the signature of real pricing-independent demand and confirms the brand moat rather than a discount-driven sugar high." Same fact, transformed. The observation reports the number; the insight tells the reader what to believe because of the number. The table below shows the move on lines you will actually write:
Advertisement
| Observation (what beginners write) | So what? (the insight a professional delivers) |
|---|---|
| "Gross margin was 42%." | "Gross margin held at 42% through a raw-material spike that cut peers 300–400bp — evidence of pass-through pricing power, i.e., the moat paying rent." |
| "The company opened 60 new stores." | "New-store productivity (revenue/sq ft of the FY24 vintage) matched the mature base within a year — the reinvestment runway is still earning the hurdle, so growth is creating value, not buying it." |
| "Promoter pledge is 0%." | "Zero promoter pledge on a promoter-controlled company removes the single most common Indian value-extraction and forced-sale risk — one fewer way for this to go to zero." |
| "The reverse DCF implies 15% growth for a decade." | "The price already assumes 15% growth for a decade; our variant claim is that the still-unorganized market makes that more likely than the market's implied fade — the edge is runway length, not quality." |
Notice that the insight always ends up pointing at one of the five spine questions or at the decision. That is the tell that you have reached "so what": an insight is actionable or belief-changing; an observation is merely true. A memo made only of observations is a homework assignment; a memo made of insights is analysis.
What is throat-clearing, and how do you cut it?
The second failure is throat-clearing: sentences that occupy space before the point arrives. "In this memo, I will discuss the investment merits of Titan Company." "It is important to first provide some background before turning to our recommendation." "There are many factors to consider when evaluating this business." Every one of these is a sentence the reader must pay for and receives nothing from; each delays the conclusion the pyramid says should come first. Cut them without mercy. The reader knows they are reading a memo about Titan; they do not need to be told that a memo about Titan is coming. Related sins to hunt and kill: hedging clouds ("it could perhaps be argued that there may be some risk that…": say what you believe and mark your confidence), passive evasions ("mistakes were made": name the agent), and adjective inflation ("a truly exceptional, world-class, best-in-breed franchise": the adjectives are doing the work the evidence should do, and a good reader discounts every one). The governing question for every sentence is Minto's and Strunk-and-White's at once: does this earn its place? If a sentence can be deleted without loss of meaning, deleting it is a gain in meaning, because it raises the signal of everything around it.
How do you use the so what test as an editing pass?
Put the two moves together and you have the editing pass that turns a draft into a memo: lead with the answer (Pyramid), make every retained sentence an insight not an observation (so-what), and delete everything that does neither (throat-clearing). Most first drafts shrink by a third under this pass and improve at every line. The compression is not stylistic tidying; it is the same Feynman instrument again: the sentences you cannot make into insights are the places your analysis is thin, and cutting them honestly shows you where more work is owed.