How to Compare Stocks Without Creating False Precision

The short answer

A comparable-company table is a model, not raw fact. First decide why the businesses are peers. Then label each fiscal period, restate currencies with a dated exchange rate when necessary, reconcile metric definitions, and separate structural business-model differences from temporary cycle effects. Use ranges and scenarios when the evidence cannot support a precise adjustment.

The worked example uses Alphabet's calendar-year 2024 Form 10-K and Microsoft's fiscal-year 2024 Form 10-K, which ended June 30, 2024. Both report in USD millions under U.S. GAAP, but their periods differ by six months and Microsoft completed its Activision Blizzard acquisition during its fiscal year. Those differences prevent the table from being a perfectly synchronized competition.

1. Write the peer-selection thesis

Alphabet and Microsoft are reasonable peers for questions about large-scale digital platforms, cloud infrastructure, advertising exposure, software ecosystems, and AI investment. They are weak peers if the question assumes identical revenue models. Alphabet's filing separates Google Services, Google Cloud, and Other Bets; Microsoft reports Productivity and Business Processes, Intelligent Cloud, and More Personal Computing. Segment names and allocations are not interchangeable.

A defensible peer group states both inclusion and exclusion rules. Similar market capitalization is not enough. Record customer type, pricing model, asset ownership, regulatory capital, acquisition intensity, geography, and cyclicality. If one company is present only because a data vendor assigns the same industry label, remove it.

2. Build an as-reported table with visible period labels

The formulas are reproducible. Alphabet's operating margin is $112,390m / $350,018m = 32.1%; Microsoft's is $109,433m / $245,122m = 44.6%. Simple free-cash-flow margins are $72,764m / $350,018m = 20.8% and $74,071m / $245,122m = 30.2%. These are useful observations, not proof that Microsoft is universally “better.”

3. Choose a period-alignment method before ranking

There are three honest choices. Keep each latest annual period and show the exact end dates, as above. Build synchronized trailing-twelve-month figures from quarterly filings. Or compare common calendar quarters and accept shorter-term seasonality. Do not silently mix a December annual report with a June annual report and label both “2024” as if they covered the same demand, interest-rate, and currency environment.

A trailing-twelve-month reconstruction needs the latest 10-K plus subsequent 10-Q data and must handle year-to-date cash-flow statements correctly. It also needs acquisition and discontinued-operation adjustments. This page deliberately does not invent synchronized figures from annual reports alone.

4. Normalize currency without erasing economics

Both examples report in USD, so no conversion is required. When currencies differ, store the native values and the converted values. Use average rates for period income and cash-flow items, closing rates for point-in-time balance-sheet items, and a clearly dated market rate for valuation. Then disclose whether reported growth or constant-currency growth is being compared.

Currency conversion makes units comparable; it does not make business exposure identical. A company may report in USD while earning substantial foreign revenue and incurring costs in other currencies. Read the sensitivity and hedging disclosures rather than assuming reporting currency equals economic currency.

5. Reconcile accounting and capital-allocation definitions

Microsoft's filing reports $69,132 million of acquisition cash outflow, largely reflecting Activision Blizzard. Alphabet's displayed simple FCF does not deduct acquisitions either. A comparison focused on recurring organic reinvestment may show FCF before acquisitions, while a stewardship review must also show acquisition spending and resulting goodwill or intangible assets.

Stock-based compensation was $22,785 million for Alphabet and $10,734 million for Microsoft in their displayed cash-flow reconciliations. The amounts were 6.5% and 4.4% of revenue, respectively. Because SBC is added back in operating cash flow, an analyst should present the reported FCF and a clearly labeled owner-cost scenario rather than allowing different award intensity to disappear.

6. Replace point rankings with an evidence range

Use reported, normalized, downside, and upside cases. A normalized case can remove a disclosed one-time item, but it should not remove recurring “one-time” restructuring every year. A cyclical case should move revenue, margin, working capital, and capital expenditure coherently. Show which conclusion survives reasonable assumptions; that is more decision-useful than declaring a winner from a 0.3-point ratio gap.