How I Read Annual Reports

The 10-K is one of the most information-dense documents a public company produces. It's also designed to be as unrevealing as possible within the constraints of what regulators require. Reading it well is a skill that takes practice.

Here's how I approach it.

Read in Reverse

Most people start at the beginning — the business description, the letter to shareholders, the strategic overview. This is where the marketing is. The company controls this narrative completely.

Start at the back.

The financial statements, footnotes, and auditor report are where legally required disclosures live. The MD&A (Management Discussion and Analysis) is the transition zone — partially controlled narrative, partially required disclosure.

By starting at the back, you build an independent view of the financial reality before you encounter management's frame for it.

The Footnotes Are the Story

The financial statements present a clean summary. The footnotes contain the assumptions, the accounting policy choices, and the items management would prefer you not focus on.

Things I specifically look for in footnotes:

• Revenue recognition policies: How does the company recognize revenue? Are there changes from prior periods?
• Deferred revenue trends: Growing deferred revenue in a SaaS company is a leading indicator; declining is a warning sign.
• Related party transactions: Any transactions with parties related to management or major shareholders.
• Contingent liabilities: Lawsuits, regulatory investigations, warranty obligations.
• Pension and post-retirement obligations: Often significant and often buried.

The Auditor Letter

Two things to check immediately: the auditor's name, and any "emphasis of matter" paragraphs.

A Big Four auditor on a small company is a mild positive signal. A small auditor on a large company is worth investigating. An auditor change mid-stream always warrants understanding why.

Any language beyond the standard audit opinion format is a flag. Auditors are conservative by nature — if they're raising something, it's because they thought it warranted raising.

MD&A: Read for What Isn't There

The MD&A section is where management explains results. The analysis should be: does this explanation match what the financial statements show?

Specific tests:

• When revenue grows, does the explanation attribute it to volume, price, or mix? If price, is that sustainable?
• When margins compress, is the explanation credible given what the footnotes show about cost structure?
• If a metric the company highlighted last year has disappeared, why?

The absence of a previously prominent metric is often more informative than what's present.

Segment Disclosure

If the company has multiple operating segments, the segment disclosure in the notes is frequently the most valuable section of the entire document. Consolidated results can mask divergent performance across business units.

Look for: which segment is growing, which is contracting, which has better margins, and whether capital is being allocated toward or away from the high-return segments. Companies that cross-subsidize declining segments with profitable ones often obscure this in headline reporting.

The Proxy Statement as Companion Document

The 10-K doesn't tell you much about how management is compensated. The proxy statement does. Incentive structures predict behavior.

If the CEO's bonus is tied to revenue growth with no profitability component, expect revenue-maximizing decisions regardless of economics. If long-term equity is tied to total shareholder return, watch for financial engineering in the service of short-term stock performance.

Build the Model After, Not Before

The most common mistake in financial analysis is building a spreadsheet model before reading the 10-K carefully. The model becomes the frame, and you read the document to populate cells rather than to understand the business.

Read the document first. Form a qualitative view of the business: how does it create value, where does it earn returns, what are the key risks? Then build a model that tests whether the quantitative history is consistent with that qualitative story.

Discrepancies between the story and the numbers are where the most interesting analysis begins.