Full fundamental analysis of a company, end to end

The answer

A full fundamental analysis is not a formula. It is 9 questions asked in a fixed order, where the early answers decide whether the later questions are worth asking at all.

It ends in 1 page you write yourself: what the business is, what has to stay true, what the price assumes, your verdict, and — the part almost nobody writes — the specific things that would tell you that you were wrong.

Why this costs you money

You have read everything in this cluster. You know what a moat is, how to read a cash flow statement, what net interest margin means, how to reverse a discounted cash flow model. And you still buy the way you always did, because the knowledge is in 16 separate pieces and a decision needs 1 answer.

Three specific losses come from having the pieces but not the order.

You start with valuation. Somebody mentions a company. You check the P/E. It looks reasonable, so you look further, and everything you look at afterwards is read in the light of that first favourable impression. Valuation is the last question, not the first, because a cheap price on accounts you cannot trust is not cheap. It is a smaller number attached to fiction.

You never write anything down. So 8 months later you cannot remember why you bought. The share has fallen 30% and you have no way to tell whether your reason has broken or whether the price has simply moved. Without a written thesis, every fall looks like an opportunity and every rise looks like confirmation. You will hold the losers and sell the winners, and you will not know you are doing it.

You have no exit condition. This is the expensive one. If you cannot state in advance what evidence would change your mind, then no evidence will, and you have not made an investment decision. You have adopted a company. People hold shares for 6 years through deteriorating returns on capital, rising receivables and 2 auditor changes, because at no point did they write down what would be enough.

How it works

The order, and why it is this order

The questions are arranged so that the cheapest disqualifying questions come first. Each stage can end the analysis. That is the point: most companies should be rejected quickly, and the time saved is spent on the few that survive.

Question 1. What does this company sell, to whom, and how does it get paid?

Before any number. Read the business description in the annual report and the segment note. Write 3 sentences: what the product is, who buys it, and what the buyer's alternative is. Then find revenue by segment and by geography, and the share of revenue from the largest customers.

Stop condition: if you cannot write those 3 sentences from the company's own filing, stop. You are not going to analyse your way out of not understanding the business.

Question 2. Which family is this, and therefore which metrics apply?

Lender, capital-heavy producer, or asset-light service business. This decides every ratio you will use for the rest of the analysis. A bank gets price to book against return on equity, net interest margin, cost-to-income, gross and net non-performing assets, provision coverage and capital adequacy. A factory gets enterprise value multiples, capacity utilisation and net debt to EBITDA. A software business gets growth, gross margin, retention and cash conversion.

Get this wrong and everything after it is wrong.

Question 3. What weather does this business live in?

The top-down question. What is the company sensitive to: interest rates, commodity prices, the exchange rate, a regulator, a government subsidy, a single end market. You are not forecasting the economy. You are identifying which 2 or 3 external variables move this company's profits, so that later you know what to watch.

Question 4. What is the industry structure, and does this company have an advantage inside it?

How many competitors. Whether prices are set by the industry or by the company. Market share and its direction over 5 years. What stops a new entrant. Whether customers can switch easily and what it costs them.

The evidence for a real advantage is financial and it is simple: return on capital employed that has stayed high for 8 to 10 years, through at least 1 bad year. Two good years is a cycle. Ten is a structure.

Question 5. Can I trust the accounts?

This is the gate. Everything after it assumes the answer is yes.

Run 3 checks. Cumulative net profit against cumulative cash from operations over 5 years. The red flag sweep: auditor changes, promoter or insider pledging, and the interest yield earned on the reported cash balance. And receivable days over 5 years.

Stop condition: if the auditor resigned mid-term, or 3 red flags are present at once, stop. Do not proceed to valuation. There is no price that compensates for accounts that may not describe the company.

Question 6. How good is the business, in numbers?

Return on capital employed for 10 years, written as a list, not an average. Gross and operating margin over the same period. Asset turnover. Working capital days. For a lender, return on equity and return on assets over 10 years instead.

You are looking at 2 things: the level, and the stability. A business earning 15% every year for 10 years is better than one averaging 18% with a range from minus 5% to 40%.

Question 7. How strong is the balance sheet if business gets bad?

For most companies: net debt to EBITDA, interest coverage, the maturity profile of the debt, contingent liabilities and guarantees given. For a lender: capital adequacy, tier 1, provision coverage, and the maturity match between borrowings and loans.

The question is not "is there debt". It is "what happens to this company if revenue falls 25% for 2 years".

Question 8. Where does growth come from, and who is funding it?

Split revenue growth into volume, price, mix and acquisitions. Compare actual growth over 5 years with the sustainable growth rate, which is return on equity multiplied by the proportion of profit retained. If the company grew faster than it could self-fund, find out who paid: lenders, or new shareholders through dilution.

Question 9. Who runs this company, and for whom?

The capital allocation record over 10 years. The related-party transactions note. The pledge percentage. The auditor's fee split. And the 5-year promise check: open the annual report from 5 years ago, list the specific commitments in the chairman's letter, and mark each one against what happened.

Question 10. What does today's price assume?

Only now. Two calculations.

The multiple appropriate to the family, compared with the company's own 10-year range and with 2 direct competitors. And the reverse question: implied growth = your required return − the free cash flow yield. State the answer as a sentence. "At this price the market assumes free cash flow grows about 7% a year forever." Then test that against the company's own last 10 years, its industry's growth, and long-run nominal GDP growth.

The one-page verdict

Write it. Not in your head. One page, and it has 8 parts.

1. The business, in 3 sentences. What it sells, to whom, and why they buy it rather than the alternative.
2. The family and the metric set you used.
3. The 3 things that must keep being true for this to work. Be specific and observable. "Volume growth stays above 6%" is usable. "The company stays well-managed" is not.
4. Accounts: the 5-year cash conversion ratio, receivable day trend, red flag count.
5. Quality: the 10-year return on capital range, in 1 line.
6. Balance sheet: the 1 number that matters for this family, and what happens in a 25% revenue fall.
7. Price: the multiple, its 10-year range, and the implied growth rate. Then 1 sentence: do you believe it?
8. Verdict: own, do not own, or own smaller. And the position size, with the reason.

The part nobody writes

Under the verdict, a heading: what would make me change my mind.

Three conditions. Each must be specific, observable in a public filing, and capable of occurring. Then a review date.

Examples of the right shape:

• "Return on capital employed falls below 12% for 2 consecutive years."
• "Receivable days rise above 110."
• "Promoter pledge exceeds 25% of holding."
• "Volume growth turns negative for 3 consecutive quarters."
• "The auditor changes for any reason other than scheduled rotation."
• "Net debt to EBITDA rises above 3."

Examples of the wrong shape: "if the story changes", "if the market turns", "if the price falls 30%". A price fall is not evidence about the business. It is the thing you are trying to interpret.

Write these before you buy, because after you buy you will not be able to write them honestly.

What it tells you, and what it does not

This process tells you whether you understand what you own, what you are paying, and what you are assuming. That is the whole of what analysis can deliver.

It does not tell you what the share will do. A company can pass every question and fall 40% because interest rates rose. It does not tell you when. And it does not scale to a large portfolio: doing this properly takes 2 to 4 hours per company, which is a real constraint and an argument for owning fewer things.

It also has a specific blind spot. The process is built on the company's own disclosures. It tests them for internal consistency, and it can detect many kinds of misstatement, but it cannot see what has not been disclosed. Question 5 is a gate, not a guarantee.

The decision rule

Ask the questions in order. Let the early ones stop you. And never let a low price make you skip question 5.

The gating rule. Questions 1, 2 and 5 can each end the analysis on their own. If you cannot describe the business, if you have the wrong metric family, or if the accounts cannot be trusted, nothing downstream is worth computing.

The conditional that matters most. A company that passes questions 1 to 9 strongly and looks expensive at question 10 is usually worth watching and occasionally worth owning anyway — unless the implied growth rate exceeds long-run nominal GDP growth, in which case the price requires something arithmetically impossible and no amount of business quality fixes it.

The reverse conditional. A company that looks cheap at question 10 and weak at questions 5, 6 or 9 is not a value opportunity. It is a company that is cheap for a reason you have already found.

Position size follows conviction and fragility, not just conviction. A company with excellent quality and 1 unexplained red flag gets a smaller position, not a rejection and not a full one.

Try this now

Five minutes, on the holding you own the most of. This is the smallest useful piece of the whole process, and it is the piece with the highest return.

1. Open a blank note. Write the name of your largest holding at the top.
2. Write 3 sentences: what this company sells, who buys it, and what those buyers would use instead. Do it from memory first. Then open the company's latest annual report and check whether you were right.
3. Write the heading what has to keep being true, and list 3 things. Make each one a number you could look up in a filing.
4. Write the heading what would make me change my mind, and list 3 conditions. Each must be specific, observable in a public filing, and possible. No price levels. No feelings.
5. Write a review date 6 months from today, and put it in your calendar.

What you should see. Two uncomfortable discoveries, and both are the point.

Step 2 is harder than it looks. Many people find they cannot describe the business of a company they have owned for 3 years without checking, and that when they check, the revenue mix is not what they assumed. If the annual report surprises you, you did not own what you thought you owned.

Step 4 is harder still. Most people's first attempt produces conditions that cannot happen or cannot be observed. Rewrite them until each one is a number in a filing. When you have 3 real conditions written down, you have converted a holding into a decision, and you have given your future self a way to be honest.

Then, when you have 3 hours, run the full 10 questions on the same company. You will already have questions 1 and the falsification list done.

Three real cases

1. Nokia, 2007 to 2013 (Finland, listed in the United States) — every statement clean, the industry question missed At the end of 2007 Nokia held roughly 40% of the global mobile phone market. Its accounts were not in question. Cash generation was strong, the balance sheet was sound, and on trailing profits it looked reasonably priced or cheap in most years that followed. An analysis that ran questions 5, 6, 7 and 10 would have produced a favourable answer for several years running. Revenue fell from about €51 billion in 2007 to about €12.7 billion in 2013, and market value fell more than 90% from its 2007 level. The question that mattered was question 4: whether the advantage was durable against a change in what a phone was. No ratio answers that. It is why question 4 comes before questions 6 and 10, and why an answer at question 10 cannot rescue a wrong answer at question 4.

2. Eastman Kodak, January 2012 (United States) — the moat that was real and then was not Kodak dominated photographic film for most of the twentieth century and held genuine technological advantages, including early digital camera patents. Its returns on capital had been high for decades, which is exactly the evidence question 6 looks for. The industry it dominated shrank faster than it could convert. The company filed for Chapter 11 bankruptcy protection on 19 January

1. The lesson for the order of questions: 10 years of high returns on

capital is evidence of a past structure, not a guarantee of a future one, and question 4 must ask specifically what would have to happen for the advantage to stop mattering.

3. Jet Airways, April 2019 (India) — the balance sheet question, asked too late Jet Airways was among India's largest private airlines. Its financial position deteriorated over several years, with mounting debt, negative net worth in reported accounts and delayed payments to lessors, staff and suppliers. It suspended all operations on 17 April 2019 after failing to secure funding, and was subsequently admitted to insolvency proceedings. Airlines are a capital-heavy business in an industry with weak pricing power, high fixed costs and fuel and currency exposure. Questions 3, 4 and 7 together — the weather, the industry structure and what happens if revenue falls — would have produced a clear answer years before the suspension. The information was in the published accounts throughout.

The question that resolves it

A novice finishes researching a company and asks: should I buy this?

An expert finishes the same research and asks: what would have to happen for me to conclude I was wrong, and would I actually notice it?

The first question has an answer that feels final and is usually just the last thing you read. The second creates a record, sets a review date, and makes the next 5 years of holding into a series of decisions rather than a single act of faith.

What would make this wrong

If this process reliably produced good investments, everybody following it would beat the market. They will not. The process is mostly a defence: it removes categories of avoidable error. Removing errors improves results. It does not create insight, and insight is what produces returns above average.

The specific limits, stated honestly.

It is slow. Two to 4 hours per company is a genuine cost. A reader who can give investing 3 hours a month can follow perhaps 10 companies properly. That is an argument for a concentrated portfolio or for index funds, and both are reasonable conclusions to reach from this article.

It is biased toward the measurable. Everything in questions 5 through 10 is computable. Questions 1 and 4 are the ones that decided the outcome in all 3 cases above, and they are judgement. A process that feels rigorous because it produces numbers will systematically underweight the questions that matter most.

It cannot see undisclosed information, and it gives no timing at all. A correct verdict can be 5 years early, which is indistinguishable from wrong while you wait.

And the falsification list has its own failure mode. Written conditions can be quietly renegotiated. When receivable days cross the level you wrote down, the temptation is to decide the level was too strict. The defence is to write the date next to the condition and to review it with somebody else, or at least to record your reasoning at the time so that you can see later what you did.

Carry this

• Ask the questions in order. Business, family, weather, industry, accounts, quality, balance sheet, growth, management, price. Price is last.
• Question 5 is a gate. If the accounts cannot be trusted, there is no price that makes it worth continuing.
• End with 1 page, and put 3 falsifiable conditions and a review date at the bottom. If you cannot say what would change your mind, you have not made a decision.

Knowledge check

Related

• Intrinsic value and discounted cash flow, made simple
• Earnings quality and accounting red flags
• Why you cannot value a bank like a technology company
• Management, governance and business quality
• ← Fundamental analysis, end to end