The two-hour first pass

You have a company name, an internet connection, its latest annual report, and two hours. You have never looked at it before. At the end of the two hours you need an honest, defensible first view — not a valuation, not a decision to own it, but a verdict on whether the company is worth more of your time. How do you spend the two hours?

This is the first module of Part Ten, and it is the whole book applied at once. Everything the earlier parts built — reading the statements, reconciling cash to profit, scanning for forensic flags, judging the promoter, placing the business against peers, decomposing growth, seeing what the price already assumes — collapses here into a single, repeatable sitting. The skill is no longer any one of those readings. It is doing all of them, in the right order, under a clock, and stopping with a clear head.

The reason to have a fixed routine is that the alternative — reading a company in whatever order your curiosity wanders — reliably produces an unbalanced view. You fall for the first interesting thing, spend an hour on it, and never reach the dimension that would have killed the idea in five minutes. A first pass is a routine for breadth before depth: it guarantees that every dimension gets a look before any one gets your love, because the dimensions you skip are exactly the ones a good story is hiding behind.

The reason for the clock is time-boxing — a hard limit on each stage so that effort is allocated by importance and diminishing returns rather than by how absorbing the material happens to be. Research has steeply diminishing returns: the first fifteen minutes on the promoter tell you most of what a first pass needs; the next two hours on the same promoter tell you very little more and cost you every other dimension. The time box is what turns "I could read this forever" into "I have looked at all seven things and can now decide whether to look harder."

This is why the module sits where it does. Part Nine closed on re-underwriting a holding you already own; this part turns outward to the blank page — the company you have never seen — and the two-hour pass is the funnel that decides which blank pages ever earn the days a full underwrite costs.

The routine is seven passes, run in order, each time-boxed, each answering one question and drawing on the part of the book that taught you how. The order is deliberate: it moves from what the business is through whether the numbers are real and whose company it is to what the price already assumes — the cheapest killers first, the price last, because the price only matters once the business has survived the earlier passes.

Read the passes as a chain, each handing the next a sharper question:

1. What does it do — 15 minutes (Parts One–Two). Before a single ratio, answer plainly: how does this company turn effort into money, and which sector's economics govern it? A lender, a developer, a QSR chain and a contractor are read with different instruments (104), so this pass decides which instruments the next six passes will use. If you cannot explain the business in two sentences, that is itself a finding.
2. The three statements — 20 minutes (Parts One, Three). A quick read of the P&L, balance sheet and cash flow together, aimed at two things: does profit turn into operating cash over time (the cash-conversion question), and how much leverage sits behind the returns? Cash quality and solvency are what the later, more interesting passes all rest on.
3. Forensic quick-scan — 20 minutes (Part Four). Not a full forensic audit — a scan for the handful of flags that cluster: receivables outrunning sales, cash far below profit, related-party dependence, contingent liabilities large against net worth, an auditor signal. One flag is a question; a cluster is a verdict. This is the pass most likely to end the whole exercise early, which is why it comes before you have fallen for anything.
4. Promoter & capital allocation — 20 minutes (Part Five). Whose company is it, and where does the cash actually go? A quick promoter read — holding, pledge, related-party pattern, the decade's capital-allocation record. In India the controller's quality is often the first filter, ahead of the business (073).
5. Position & price — 15 minutes (Part Six). Is there any durable edge, and is the business cheap or dear against a real peer set — peers that share its economic model, not merely its sector label? A rough relative read, not a valuation; the aim is to know whether the price is in a sane zone.
6. Growth & leading indicators — 15 minutes (Parts Seven–Eight). What actually drives the growth — volume, price, mix, new capacity — and what leading indicators would signal it turning before the P&L does? Enough to judge whether the growth is durable and whether you would know early if it broke.
7. What's priced in — 15 minutes (Part Nine). Finally, the price as a claim: what future does the current price already assume, and — the sentence that ends the pass — what specific fact would change my mind? If the price embeds perfection, even a wonderful business is not a first-pass keep (118).

Laid out together, the routine shows its logic — the time box on each pass roughly tracks how much a first view depends on it, and every pass carries a kill signal: the thing that, if you find it, ends the pass (and sometimes the whole exercise) then and there. Reading down the kill-signal column is itself a fast way to run the pass.

Walk it live. Take a composite mid-cap specialty-chemicals maker we will call Meridian Chemicals illustrative. Pass 1 (12 minutes): it makes intermediates for agrochemical clients, so this is a capital-heavy, cyclical, working-capital-hungry business — read it with those instruments. Pass 2 (20 minutes): revenue up strongly, but five-year cash conversion is only 0.6, and receivable days have crept from 70 to 105; net debt is rising to fund it. [illustrative] A yellow light, not yet a kill. Pass 3 (20 minutes): the receivable build reconciles to a genuine shift into export clients on longer terms — one flag, explained, not a cluster. Pass 4 (18 minutes): promoter holding is high and unpledged, related-party dealings are small, and a decade of capex has actually earned its return — the strongest pass. Pass 5 (15 minutes): it trades in line with genuine peers, no obvious edge but no obvious excess. Pass 6 (13 minutes): growth is real volume from a new plant now ramping, with capacity utilisation as the leading indicator to watch. Pass 7 (12 minutes): the price assumes the cycle stays benign and the new plant fills — and the fact that would change your mind is a stall in that plant's utilisation. Total: under two hours, and a clear-headed verdict without a single figure of your own invented as a valuation.

The output of the first pass is one of exactly three verdicts, and none of them is buy:

• Reject. Something in the seven passes is disqualifying — a forensic cluster, a governance veto, a business you cannot understand, a price with no room. The company goes on the reject pile, and the discipline is that most companies belong there. A fast, well-reasoned reject is the routine working, not failing — it is the pass earning back the days a full underwrite would have wasted.
• Dig deeper. The passes read well except for one specific, resolvable question — the arm's-length nature of a related-party line, the durability of a margin, the reality of an order book. Dig-deeper is a real verdict, not a soft keep: it means no view is possible until that one named question is answered, and it hands you the exact next task.
• Keep watching. The company is readable, interesting and clean, but there is no reason to act now — the price is fair rather than compelling, or the thesis needs a catalyst that has not appeared. It goes on a watch list with a note on what would move it to a full underwrite.

Notice what the routine deliberately does not produce. It does not produce a target price, a buy, or a position size — those need the full underwrite that a keep or a resolved dig-deeper earns, not the two-hour pass. The value of the first pass is the repeatable process, not any single verdict: run honestly across many companies, it reliably surfaces the few worth days of work and rejects the many that are not, and that sorting — not being right about any one name — is what compounds.

A first pass is a filter, and it has the honest limits of a filter. It cannot value the company. It can tell you the price is sane, insane, or worth checking; it cannot tell you what the business is worth, because that needs the modelling a keep verdict earns and the pass explicitly refuses.

It cannot confirm a fraud, or clear one. Two hours is enough to see flags cluster, which is a reason to reject or dig deeper; it is nowhere near enough to prove manipulation or to certify accounts as clean. A first pass that finds nothing alarming has found nothing alarming in a fast scan — it has not audited the company, and reading a quiet scan as a clean bill of health is a category error.

It cannot distinguish a keep from a dig-deeper by feel. The temptation, when six passes are lovely and one question nags, is to round the nag away and call it a keep. The routine's integrity depends on refusing that: a live, specific, unresolved question is a dig-deeper, and pretending otherwise smuggles a decision past the one pass that would have stopped it.

And it cannot replace the depth it triages toward. The pass is a beginning, not an end — its whole purpose is to decide what deserves the end. Treating a two-hour keep as a finished investment case is the routine misused as a conclusion rather than a filter.

The first trap is not stopping — the rabbit hole. A pass gets interesting, the clock is ignored, and two hours become an afternoon spent on one dimension while six go unread. Time-boxing exists precisely because the material is absorbing; the reader who cannot leave Pass 2 has not been thorough, they have been captured, and they end with the unbalanced, over-confident view the routine was built to prevent.

The second is falling for Pass 1 and reading the rest to confirm it. A good story in the first fifteen minutes sets the frame, and every later pass gets read as supporting evidence rather than as an independent test. The antidote is to run the kill signals as genuine attempts to reject — to read Passes 3, 4 and 7 looking for the reason to stop, not for permission to continue.

The third is treating dig-deeper as keep. The three-verdict system only works if the middle verdict is respected as a hard gate — one named question that must be answered before any view exists. Collapsing it into "probably fine" is how a real, unresolved doubt gets waved through under the costume of caution.

The fourth is mistaking a fast reject for rigour, or a slow read for it. A first pass can reject well in twenty minutes and can also reject lazily in five, dismissing a good company on a surface flag never checked. The skill is neither speed nor slowness but calibration: spending the time box where the verdict actually turns, and being willing to move a hasty reject to dig-deeper when the flag might have an innocent explanation you did not pause to find.