You do not read a company front-to-back like a novel. You assemble a stack of documents in a deliberate order, because the order is itself an analytical tool: you build the bull case from the company's own mouth first (so you understand what it's trying to be), then you get the numbers (so you can test the claim), then you find the adversary (so you can try to break it). Four layers.
What does the company say about itself?
Layer 1: the company's own "what we do." Start where the company explains itself, because you cannot judge a number you cannot narrate.
- US: the 10-K, Item 1 (Business) and Item 1A (Risk Factors). Item 1 is the company's own map of itself: segments, customers, competition, how it makes money. Item 1A is a legally-driven risk list; you mine it for the specific (a named customer at 22% of revenue; a patent cliff in 2027) and ignore the boilerplate. Read Item 1 in full (it is usually 5–15 pages); hunt Item 1A rather than read it.
- India: the MD&A and the segment note. India has no Item 1; the equivalent "what we do and how it went" lives in the Management Discussion & Analysis (industry structure, segment performance, outlook) and the segment note in the financial statements (revenue and result by segment, which is where you find the profit engine). Read them together.
- Often highest-signal of all: the latest investor presentation. Companies put their cleanest self-explanation in the deck they show investors: the business model on one slide, the KPIs they want you to watch, the strategy in bullets. It is also the most curated document they produce, so hold it in both hands at once, highest signal for understanding the business and most advocacy-laden for judging it. Read it first to orient; believe none of the numbers until Layer 3 confirms them.
Advertisement
What do the earnings calls tell you?
Layer 2: the latest one or two earnings-call transcripts. This is the highest signal per minute of anything you will read. A transcript gives you three things a filing cannot:
- Framing: what management chooses to talk about first, and which metrics they lead with (the metric they lead with is the metric they're proud of; the one that vanished is the one that went bad).
- What the analysts push on: the Q&A is a free panel of professionals who cover this company full-time, and their questions cluster around exactly the soft spots you'd otherwise take days to find. If four analysts ask about pricing, pricing is the issue.
- Tone: do they answer straight or dodge? Own a bad quarter or blame the weather? Match this quarter's story to last quarter's promise? You are reading for candor as much as content.
Read the prepared remarks fast (they are scripted advocacy) and the Q&A slowly. The Q&A is where the truth leaks. One or two calls now; the deep dive will listen to 4–8 across good and bad quarters (M8.04).
What do five to ten years of numbers show?
Layer 3: five to ten years of numbers on one screen. Now, and only now, get the quantitative history: revenue, margins, ROCE/ROIC, cash flow (CFO and FCF), debt/net-debt, and share count, ten years wide if you can. Do not rebuild it by hand at this stage. You earned the right to use the aggregator by spreading a dozen filings by hand in Phases 1–2; the screener is now a speed tool, not a crutch. (The deep dive rebuilds it by hand anyway, M8.03; the teardown does not.) Where to pull it, free or cheap:
- India: screener.in (free; the 10-year data sheet, ratios, and a downloadable Excel are the analyst's default), Tijori, Trendlyne.
- US: stockanalysis.com (free, 10 years), TIKR, Koyfin (free tiers), wisesheets.
Your eye is doing three things as the ten years load. (1) Direction: is revenue growth accelerating or fading, and are margins expanding or eroding? (2) ROIC/ROCE level and stability: is this a good business, and has the spread persisted through a cycle (Q2)? (3) Cash reality: does CFO track PAT, is FCF positive, is share count rising or falling? Thirty seconds here answers "good business?" better than an hour of prose.
Who argues against the company?
Layer 4: one competitor's deck and one independent or bear note. Finally, borrow two outside maps:
- One competitor's investor presentation. It tells you the shared industry vocabulary, the KPIs the sector actually competes on, and (by contrast) what your company does differently.
- One independent or bear note: a short-seller report, a skeptical write-up (Value Investors Club, a Substack, ValuePickr threads for India), or the "bear case" section of any broker note. Its job is to hand you, in one document, the strongest argument against the company, the disconfirming view you'd otherwise have to construct alone (Phase 6: you steelman the bear).
The discipline for Layer 4: borrow the map, then verify every claim against the primary filing. A bear note is a hypothesis generator, not an authority. It points your Ctrl-F; it does not get to be quoted. The same skepticism you apply to the company's deck, you apply to its critic's. You borrow questions from both; you take answers only from the filings.
Why does the order matter?
Why this order and not another? Because the sequence mirrors how conviction should be built: understand what it's trying to be → see whether the numbers agree → try to break the story. Read the bear first and you anchor on a thesis before you understand the business; read the numbers first and you have ratios with no narrative to hang them on. Company's story, then numbers, then adversary. Always.