How do you compare two companies fairly?

You now have a toolkit — margins, returns, liquidity, leverage, turnover, DuPont. The danger of a toolkit is misusing it: comparing numbers that were never meant to sit side by side. Fair comparison is a skill in itself, and it's what turns ratios into real judgement.

The three fair comparisons

A single ratio means nothing until you anchor it against one of these:

• The company vs itself over time (trend). Is margin, ROE or turnover rising or falling across the last three to five years? A company is its own fairest benchmark — same business, same accounting, just a different year. Trends reveal direction that a single snapshot hides.
• The company vs direct peers (cross-section). Same industry, ideally similar size and region. A retailer's 3% margin is judged against other retailers, never against a software firm.
• The company vs the industry norm. A rough sense of "normal" for the sector puts an outlier in context — is this business unusually strong, unusually weak, or simply typical?

The traps that make comparisons lie

• Different industries. The single most common error — comparing a bank's leverage to a software firm's, or a grocer's margin to a luxury brand's. The numbers are correct and the comparison is meaningless.
• Different accounting choices. Two honest companies can report differently depending on how they value inventory, depreciate assets, or treat leases. Big differences sometimes come from the rules chosen, not the business.
• One-off items. A single asset sale, a legal settlement, a tax quirk can distort a year. That's why analysts look at several years, not one, and watch for numbers that don't repeat.
• Different year-ends and currencies. Comparing a figure from one fiscal calendar or currency to another, unadjusted, quietly injects noise.

The mindset

Ratios narrow uncertainty; they don't remove it. The professional stance is comparative and humble: "relative to its own history and its closest peers, this company's profitability has improved while its leverage has crept up" — a careful observation, never a prediction or a verdict. The next unit puts several ratios together into a full health read, and then confronts the hardest problem of all: what to do when the numbers themselves are being managed.

In the data

Two of the traps above can be read straight off a company's profile: the currency its statements are filed in, and the month its fiscal year ends. Apple's, then Walmart's:

A September year-end against a January year-end is not a like-for-like year, however similar the two businesses are. One more trap sits on the peer side: a screener lists the companies that exist today, so a peer set built this way cannot show you the group as it stood on some past date, including the companies that have since been bought or gone bust.

Try it now

Start with the unfair comparison, so you can feel why it is unfair:

1. Rank the two on profit margin, then on return on equity. The ranking flips or narrows depending on which ratio you chose — and neither company is in the other's industry, so neither ranking means anything yet.
2. Check the two profile traps from the tables under "In the data" before comparing anything: the reporting currency and the fiscal year end of each company.
3. Now set up a fair one. The screener builds the peer pool, here narrowed to one industry and US listings:

Take the top two. They are not the same size, and one of them also sells snacks; say which of those two differences you would expect to move a ratio, and in which direction. 4. Compute the same three ratios for both (net margin, ROE, debt-to-equity) from the latest year, then net margin for all three years:

Does the snapshot agree with the trend, or does the direction change the story? Comparing fairly is the difference between analysis and noise.