The Difference Between GAAP and Non-GAAP Earnings and Why It Matters
Open the earnings press release of almost any large public company and you'll find two versions of profit. There's the official number, calculated under GAAP, and then there's an adjusted number that management would prefer you focus on. The adjusted one is almost always higher, sometimes by a little and sometimes by a wide margin.
A gap between the two doesn't automatically mean someone is lying. Plenty of adjustments genuinely clarify what a business earns, and plenty of others exist to dress up a mediocre year. The useful skill, whether you're an investor reading a 10-K or an operator benchmarking a competitor, is judging each adjustment on its own merits. Here's how I think about it.
What GAAP Earnings Are
GAAP stands for Generally Accepted Accounting Principles, the standardized rulebook every US public company follows when it prepares financial statements. The Financial Accounting Standards Board writes the rules and the SEC enforces them, which is why GAAP numbers are audited, consistent across companies, and comparable over time.
GAAP earnings include everything that happened during the period. Recurring revenue, one-time gains and losses, stock-based compensation, restructuring charges, acquisition costs, impairment write-downs, all of it lands in the same number. That comprehensiveness is the point, and it's also the problem. Say a company has a perfectly normal operating year but records a $500 million write-down on an acquisition that didn't pan out. Its GAAP earnings will look far worse than what the ongoing business actually produced. The write-down is real, and it represents genuine value destruction, but it's backward-looking. It tells you about a mistake that already happened rather than about current earning power.
What Non-GAAP Earnings Are
Non-GAAP earnings are management's edit of the GAAP picture. They start with the official number, then add back or strip out items they believe don't represent the ongoing business. The usual candidates are stock-based compensation, amortization of acquired intangible assets, restructuring and severance charges, acquisition costs, litigation settlements, and impairments. Every exclusion rests on a claim that the item is non-cash, non-recurring, or unrelated to core operations. Sometimes that claim holds up, and sometimes it deserves a hard look.
The SEC requires any company that reports non-GAAP measures to publish a reconciliation showing exactly how it gets from GAAP to the adjusted figure. You'll usually find it near the back of the earnings press release, which is filed on EDGAR as an exhibit to an 8-K, or in the quarterly earnings supplement on the investor relations site. If a company promotes an adjusted number and you can't locate a clean reconciliation, treat that as a warning all by itself.
Adjustments That Usually Make Sense
Some add-backs genuinely help you understand the business.
Amortization of acquired intangibles. When one company buys another, the acquirer records intangible assets like customer relationships, technology, and brand names, then amortizes them over years. That amortization is a legitimate GAAP expense, but it doesn't consume cash and it says little about how the combined business is performing today. Adding it back gets you closer to the cash economics of the operation.
Genuinely one-time restructuring. If a company closes a plant or exits a business line as a discrete event, excluding those costs from the picture of ongoing earnings is reasonable. The qualifier carries all the weight, though. The event actually has to happen once, so watch how often supposedly one-time charges reappear.
Purchase accounting noise. Acquisitions generate accounting artifacts, deferred revenue write-downs, stepped-up asset values, and similar adjustments that distort reported earnings for years after a deal closes. Backing those out can give you a cleaner read on the underlying business.
Adjustments That Deserve Skepticism
Stock-based compensation. This is the big one, and at technology companies it's often the largest add-back on the page. The argument for excluding it is that no cash goes out the door. The argument against is stronger, in my view. Stock-based compensation pays for real work, it dilutes existing shareholders, and if the company stopped issuing shares it would have to replace that pay with cash straight out of earnings. Warren Buffett has made this point for years, asking, in effect, if stock-based compensation isn't an expense, what is it? And if it isn't a real cost, why do employees accept it as pay? Pretending it doesn't exist overstates profitability.
You can size the effect yourself. The cash flow statement shows stock-based compensation added back to operating cash flow, and the diluted share count in each 10-K tells you how much shareholders are getting diluted over time. If the share count climbs every year while adjusted earnings exclude the cost of those shares, you're being asked to ignore something expensive.
Restructuring that never ends. When restructuring charges show up year after year, they've become part of how the company operates, and excluding them every year paints an artificially rosy picture. Some industries genuinely require frequent reorganization, but that frequency is exactly why the cost belongs in the earnings you evaluate.
Litigation costs at frequently sued companies. Pharmaceutical firms, banks, and large consumer platforms face legal expenses on a fairly predictable rhythm. Labeling those costs non-recurring when they recur like clockwork is misleading.
Ordinary operating costs in a costume. Some companies have excluded items like brand launch marketing or new product development from adjusted earnings. Those are core operating expenses by any sensible definition, and stripping them out crosses the line from clarification into distortion.
A Quick Worked Example
Numbers make this concrete, so here's a hypothetical. Say a software company reports GAAP net income of $200 million. Its reconciliation adds back $150 million of stock-based compensation, $60 million of amortization of acquired intangibles, and $40 million of restructuring charges, and it lands on adjusted net income of $450 million. More than half of the adjusted profit comes from the adjustments themselves.
Go through it line by line. The intangible amortization add-back is defensible, since it's a non-cash artifact of past deals. The stock-based compensation add-back is where I push back, because that $150 million paid for engineers and salespeople who would otherwise need cash salaries. And if this is the third straight year of restructuring charges, the $40 million looks like a normal cost of doing business. My own adjusted view of this company would sit closer to $260 million, which is the GAAP figure plus the amortization. Same company, same disclosures, and a very different picture depending on which adjustments you accept.
How to Evaluate the Gap Yourself
A repeatable process beats gut feel here. This is mine.
Read the reconciliation line by line. For each item, ask four questions. Is this truly one-time? Is it unrelated to core operations? Would the business need to replace this cost with cash if it stopped issuing stock? Is it likely to recur?
Pull five years of reconciliations. Everything you need is free on EDGAR. If the same items get excluded year after year, the one-time label has stopped meaning anything, and you should treat those charges as ongoing expenses when you value the business.
Compare against peers. If one company in an industry excludes stock-based compensation and its competitors don't, the adjusted numbers aren't comparable until you normalize them. Many analysts build their own adjusted earnings for exactly this reason, accepting some exclusions and rejecting others consistently across every company they cover.
Watch the trend in the gap. When the distance between GAAP and non-GAAP earnings widens year after year, it often means the company keeps finding new things to exclude while real performance deteriorates, so a widening gap deserves extra scrutiny.
Check the proxy statement. The DEF 14A filing discloses which metrics drive executive bonuses. If incentive pay keys off adjusted EBITDA or adjusted EPS, management has a direct personal stake in making the adjustments generous, and that context should shape how much benefit of the doubt you extend. A company that pays its executives on GAAP results or free cash flow has healthier incentives around its reporting.
Check everything against free cash flow. Cash doesn't care which accounting framework you prefer. It either came in or it went out. If adjusted earnings run well above free cash flow for years, the adjusted number is almost certainly flattering economic reality. Over long stretches the two should roughly converge for a healthy business.
What the Rules Require
The rules here exist because companies abused the freedom. During the dot-com years, plenty of money-losing firms promoted pro forma earnings that excluded ordinary costs, and some looked profitable only under their own math. Sarbanes-Oxley, passed in 2002, directed the SEC to rein this in, and the result was Regulation G in 2003. It requires the reconciliation I keep mentioning, and related SEC rules require companies to give GAAP measures at least equal prominence in filings and earnings releases, so a company can't lead with the adjusted number and bury the official one.
None of that has slowed the practice down. Adjusted metrics have multiplied to the point where they can make comparison harder rather than easier, since one company's adjusted EBITDA rarely matches another's definition. The rules guarantee you a reconciliation, and the judgment about what's in it stays with you.
Practical Takeaways
Start with GAAP. It's the audited, standardized baseline that lets you compare companies without anyone's thumb on the scale. Treat non-GAAP as a supplement, and evaluate each adjustment individually instead of accepting the total at face value.
Be skeptical when the gap between the two numbers is large and growing, and doubly skeptical when the list of exclusions changes from year to year. Cross-check adjusted earnings against free cash flow over a multi-year window, because persistent divergence usually means the adjusted number is telling a story the cash doesn't support. None of this requires special tools or a terminal subscription. A press release, the reconciliation table, and twenty minutes of skeptical reading will get you most of the way there.