LESSON 29 OF 29

Why headline profit is often the wrong number

Series 6 · Lesson 1About 8 min readFree · Educational

A company reports an enormous profit. The number may be correctly reported and still tell you less about the business than its size suggests.

The useful question is not whether the headline is impressive. It is:

How representative is this number of sustainable earnings?

This is a lesson in estimating repeatable earning power, not an allegation that the accounting is wrong. It also is not a shortcut to a price target. Once you understand the profit, you still have to consider investment needs, risk and the price paid.

Two questions, not one

Start by asking where the item appears. Then ask how representative it is of future earnings.

Those questions are separate.

Where an unusual benefit appearsWhat to examineWhat the location does not establish
Below operating profit, in other incomeInvestment gains, financing items and the supporting notesThat every non-operating item is unusual or should be discarded
Within operating profitRefunds, settlements, cost reversals and their accounting treatmentThat everything included in operating profit is recurring

An investment revaluation can increase earnings without a customer buying another product. A tariff-refund benefit can reduce reported operating costs without a lasting improvement in selling merchandise.

Both can make headline profit a poor starting point for a forecast. They do so in different places on the income statement.

Ordinary interest income may recur. An unusual operating benefit may not. Operating is not a synonym for recurring; non-operating is not a synonym for worthless.

Four examples from the research archive

These examples concern specific past quarters. They are not four current investment recommendations.

Alphabet: an investment gain below the operating line

Alphabet’s Q2 2026 headline earnings included a substantial contribution from investment gains, rather than only its operating businesses. The Associated Press’s 22 July report documents that distinction. Here the example is kept qualitative; no holding-by-holding allocation is inferred.

The question is how much the customer-facing businesses earned, separately from the change in investment values. A revaluation can add asset value without representing a sustainable quarterly contribution from Search, YouTube or Cloud.

Read the original Alphabet research.

Amazon: other income is not AWS profit

For Q2 2026, Amazon reported $53.4 billion of pre-tax non-operating other income, primarily related to its Anthropic investments. That is an aggregate income-statement line, not a disclosure that every dollar was one unrealized gain. Amazon Q2 results.

Keep that line separate from AWS operating profit and the geographic segments. A valuation change in an investment does not demonstrate better package-delivery economics or a higher cloud-service margin.

Read the research on AI spending and operating evidence.

Target: a refund can affect operating profit

Target disclosed $994 million of tariff refunds received during Q2 2026, recognized as a reduction in cost of sales. The benefit was $752 million after tax, or $1.65 per diluted share. Both reported and company-adjusted EPS included it. Target Q2 earnings release, note 2.

The benefit matters economically. The forecasting mistake would be treating it as a permanent improvement in selling goods. Retain the company’s sales and traffic evidence while examining the refund separately.

Read the original consumer research.

Ross: separate the margin improvement into its parts

In the quarter ended 1 August 2026, Ross reported a 610-basis-point operating-margin improvement. Tariff refunds contributed 405 basis points; improvement excluding the benefit was 205 basis points. Diluted EPS of $2.66 included approximately $0.60 from refunds. Ross fiscal Q2 results.

One basis point is 0.01 percentage point. The margin bridge is therefore 6.10 percentage points = 4.05 from refunds + 2.05 excluding the benefit.

There was improvement beyond the refund. There was also a good reason not to extrapolate the entire reported increase. Excluding this one item does not certify that every remaining driver will recur.

Read the Amazon and Ross research.

Unusual does not mean valueless

Removing a gain from a recurring earnings forecast does not erase the asset or the money it may represent.

An investment may have substantial realizable value, subject to market risk, taxes and restrictions. A refund may improve liquidity when collected. Those can belong in an assessment of total equity value.

The mistake is counting the same value twice: valuing an investment separately while also applying a recurring earnings multiple to its unusual gain, or assuming a refund arrives afresh every quarter.

“Not a normal quarterly earning” and “has no value” are different conclusions.

Recognition is not cash collection

An unrealized investment gain raises reported earnings without sale proceeds arriving from that revaluation. Realizing an investment later is a separate event, and any proceeds still need their own tax and cash-flow analysis.

A recognized refund benefit is different, but recognition alone does not tell you when cash arrived. The company may record a receivable before collection. Read the notes and cash-flow statement before describing the benefit as cash received.

Target explicitly disclosed receipt during the quarter. Ross’s release establishes the recognized earnings benefit; its receipt timing is not established by the figures used here.

Once collected, a refund is real cash. It still is not automatically a repeatable source of cash. Conversely, a non-cash accounting gain can reflect a valuable asset.

Ask three questions: What was recognized? What was collected? What can reasonably recur?

Build a bridge, not a convenient adjusted number

A disciplined adjustment starts from the reported measure and shows exactly what changes.

Do not subtract a pre-tax gain directly from after-tax net income. Tax effects can differ, and an EPS calculation also requires the relevant diluted share count. Use disclosed reconciliations where available and label estimates where they are not.

Ross offers a simple illustration: $2.66 reported EPS − approximately $0.60 disclosed refund benefit = approximately $2.06. That removes one specified benefit. It is not a complete normalized earnings model, a new company-reported metric or an annual run rate.

Apply the same scrutiny to adverse items. Removing unusual gains while automatically accepting every “one-time” loss can distort the picture. Repeated restructuring costs or stock-based compensation do not become economically irrelevant because management excludes them from an adjusted measure.

Sustainable earnings require judgment about both income and costs. A cleaner-looking number is not necessarily a better one.

The reader’s checklist

  1. Identify the period and measure. Quarterly or annual? Operating income, net income, EPS or management-adjusted earnings?
  2. Compare operating profit with net income. For non-financial companies like these examples, a large gap is a reason to inspect other income, interest and taxes. Financial businesses require a different starting point.
  3. Inspect operating costs too. Refunds and reversals may already be inside the operating result.
  4. Read the reconciliation and notes. Record the item, amount, accounting location and reason for any adjustment.
  5. Separate recognition from collection. Check cash-flow adjustments and receivables rather than inferring cash from the profit headline.
  6. Adjust consistently. Match pre-tax with pre-tax, use the disclosed tax effect where possible, and assess unusual gains and losses symmetrically.
  7. Test repeatability across periods. Customer demand, margins and necessary costs matter more than one quarter’s label.
  8. Keep asset value and cash needs in view. Removing an item from recurring earnings does not remove its economic value or the reinvestment required by the business.

The output is a reasoned estimate with a visible reconciliation—not one magical “real profit” number.

A different test: cash left after investment

There is another question that stripping out unusual items cannot answer: how much cash remains after the business funds its investment needs?

The archive’s Meta discussion examined a period in which operating cash flow was largely absorbed by capital expenditure. That is a different mechanism from a gain lifting reported profit. It does not mean the business generated no operating cash, or that every part of its reported profit was recurring.

In a common definition, free cash flow is operating cash flow less capital expenditure; company definitions and lease treatment can differ. A business can generate substantial operating cash and retain little after investment.

That is the subject for Series 6’s next lesson. For now, keep the tests separate:

  • Earnings quality: how representative is the reported profit of sustainable earning power?
  • Cash after investment: how much operating cash remains after the relevant capital spending?

The receipt must match the claim

If the claim is stronger customer economics, look for customer revenue, operating performance and the costs needed to sustain them. If the claim is valuable investments, examine the investments. If the claim is cash available after expansion, examine cash flow and capital expenditure.

A bigger number is not a better receipt when it answers a different question.

Read the headline. Locate the benefit. Reconcile it. Then decide what belongs in the forecast.

Sources and freshness

Reviewed as of 11 September 2026. Amazon, Target and Ross examples use Q2 or fiscal Q2 2026 company disclosures. Alphabet’s qualitative example is corroborated by the Associated Press; its issuer filing was not retrieved in this review, so the archive’s $98 billion/$99 billion breakdown is not reproduced here. Archive links preserve the original research context. Figures are historical, not current forecasts.

The amounts describe reported recognition and disclosed adjustments. Cash-receipt timing is not inferred from refund recognition. The Ross EPS subtraction is an illustrative calculation, not a complete normalized result.

Educational only—my process and opinions, not personalized investment advice. Accounting classifications depend on the item and company. Investments involve risk, including loss of capital.