Stock-Based Compensation and Dilution: Read the Share Count Before the Buyback Headline

A software company announces a large buyback. Its earnings presentation excludes stock-based compensation from an adjusted profit measure. An investor then subtracts the compensation expense from cash flow and subtracts the entire buyback again to estimate an “owner” return. Another investor accepts adjusted profit and assumes the repurchase headline proves that dilution has disappeared. Both can be mixing quantities that measure different things.

The better starting point is a share-count question: what happened to an existing owner’s claim on the business, and what resources did the company use to produce that result? The answer requires more than one accounting line. It also requires a consistent valuation convention so that the same economic burden is not counted twice.

For a concrete archival example, this article reads Salesforce’s Form 10-K for the fiscal year ended January 31, 2025, filed March 5, 2025, alongside its April 2025 proxy. Those are intentionally historical documents available before the October 10, 2026 cutoff. They are not presented as Salesforce’s latest filings, and no current share price, analyst estimate, or present valuation is assumed. SEC filing index.

The first worksheet has four columns, not one verdict

Reported archival figures below use millions of dollars or shares. Salesforce fiscal 2025 Form 10-K.

This is not a scorecard declaring the stock cheap or expensive. It establishes the inputs to several distinct questions. The compensation line belongs to a recognized cost. The cash-flow lines describe resources. The share rows describe claims. The earnings numerator permits a per-share bridge.

The decrease from 971 to 962 million period-end shares is nine million shares, about 0.93%. The diluted weighted-average denominator falls by ten million, about 1.02%. Those similar percentages should not tempt the reader to treat the measures as identical. Timing, outstanding awards, and the averaging method can make them move differently in another year.

Subtracting the stated capital expenditures from operating cash flow yields 12,434 million. That is a simple cash-flow calculation using these two lines, not a claim that every analyst defines free cash flow identically. Repurchase outflow is about 63% of that amount. The proportion describes allocation of a cash measure; it does not measure the value of the shares retired.

Keep the arithmetic attached to its question. If the question is cash remaining after that repurchase outflow, subtract it. If the question is operating economics before capital distribution, treat the repurchase separately. The same line can be relevant to both analyses without belonging in both places in the same valuation formula.

Expense is recognized over a period; shares move on events

An equity award can be granted in one year, recognized as expense over a service period, and settle into shares at another time. The period’s compensation expense therefore does not directly equal shares issued during that period multiplied by the current stock price. The filing’s compensation notes provide the award terms. Salesforce fiscal 2025 compensation notes.

That timing distinction matters when the market price changes sharply. A company can recognize expense related to an earlier grant while issuing shares worth a different market amount at vesting. One number informs reported profitability; another can inform the economic value transferred at settlement. Neither should be casually substituted for the other.

All further company scenarios in this article are hypothetical arithmetic examples, not Salesforce forecasts. Consider a business granting awards valued at 300 million over three years of service. A simplified illustration recognizes 100 million annually. If those awards settle at a later price that differs from the grant-date price, the number of shares and their settlement value need not align with that year’s expense. Real accounting requires the relevant award terms and standards; the illustration isolates the clocks.

A reader can use this distinction to investigate a surprising year. Rising expense may reflect new grants, prior grant amortization, changed assumptions, or a different workforce mix. Rising issuance may reflect older awards reaching settlement. A one-year ratio cannot establish which mechanism dominates.

The useful note-taking habit is to write down both the expense period and the award event. “Expense recognized this year” and “shares delivered this year” are different observations. The equity-award rollforward, unrecognized compensation information, and vesting disclosures can help connect them, but the connection should be made from disclosed terms rather than guessed from a headline.

This also makes compensation comparisons more credible. Two businesses with the same current expense can have different future award pipelines. One may have recently granted substantial awards; the other may be near the end of a large historical grant’s recognition period. Current margins alone cannot reveal that difference.

A share bridge explains what a buyback actually accomplished

Take a hypothetical company with 100 million shares outstanding at the beginning of a year. Employees receive three million net new shares. The company repurchases four million shares. Ignoring other changes, the period ends with 99 million shares.

The gross repurchase was 4% of the opening share base. The net reduction was 1%. An owner who did not trade owns a slightly larger fraction of the business, but the company used enough cash to purchase four million shares to achieve that one-million-share net decline. Describing the entire repurchase as ownership growth obscures the issuance it offset.

The bridge is simple in concept: opening shares, plus shares entering the outstanding base, minus shares leaving it, equals closing shares. In real filings, settlement withholding, acquisitions, employee purchase plans, conversions, and treasury-stock activity can complicate the components. Treat the bridge as a reconciliation task, not a reason to infer missing components from one residual.

A residual is still informative. If a company purchases ten million shares but outstanding shares fall by only two million, investigate the other eight million. They could relate to employee equity, transaction consideration, or another disclosed event. The arithmetic identifies a question; it does not prove the answer.

There is also a difference between shares purchased and shares whose effect appears in a weighted-average earnings denominator. A purchase near year end has little time to reduce that year’s average. A similarly sized purchase early in the year has much more. Comparing repurchase dollars with the same year’s diluted denominator without considering timing can understate or overstate the apparent offset.

For investors comparing businesses, a multiyear share bridge is often more informative than a single buyback announcement. It shows whether gross repurchases repeatedly offset issuance, whether the base actually contracts, and whether exceptional transactions dominate the pattern. The result remains a history of allocation, not proof that management paid an attractive price.

A steady denominator can conceal a cash obligation

A hypothetical settlement also shows why the cash-flow statement belongs beside the share bridge. Suppose employees become entitled to two million shares, but only one and a half million enter the outstanding base because half a million are withheld under the plan’s settlement arrangements. If the company remits cash associated with that withholding, the smaller share increase does not imply that the entire settlement was economically smaller. The actual treatment must come from the award terms and cash-flow disclosures.

An investor looking only at the ending denominator can miss this channel. An investor looking only at the recognized expense can miss its timing. The reconciliation should therefore ask which award-related transactions changed outstanding shares, which used cash, and which remained potential future claims. This is a framework for reading disclosures, not an assertion that the hypothetical amounts occur in Salesforce’s filing.

For comparison, a company might repurchase one and a half million shares and report a steady outstanding base in this scenario. Calling that result a complete elimination of compensation cost would ignore both the repurchase cash and the settlement-related cash. The steady denominator is accurate; the interpretation needs the resource bridge as well.

Why “SBC divided by stock price” is only a rough thought experiment

Suppose a hypothetical business recognizes 100 million of compensation expense and its shares trade at 50. Dividing the expense by the price yields two million “equivalent shares.” That calculation may help illustrate scale, but it is not a forecast of actual issuance.

The expense can include awards with different settlement dates, conditions, and accounting values. Some awards may be forfeited; some may be settled differently. Even for restricted stock units, grant-date values and current prices can differ substantially. The denominator may also be a spot price selected on a day that has little relationship to actual repurchase execution.

The calculation becomes especially misleading when used to claim that a buyback neutralizes compensation. If shares vest when the price is 80, a 100 million repurchase buys 1.25 million shares. If shares vest when the price is 40, it buys 2.5 million. The same dollars can offset different numbers of shares depending on timing.

A better approach first observes disclosed issuance and repurchase shares, then uses dollars to analyze cost. If those disclosures do not permit a clean bridge, state the limitation. A carefully bounded estimate is useful; an estimate presented as a reconciled fact is not.

This distinction changes how an investor interprets market declines. A lower price can make a fixed-dollar repurchase more powerful in share terms. It can also alter employee compensation incentives and future grant decisions. Those are separate channels. One does not automatically compensate for the other.

The arithmetic shortcut therefore belongs in a sensitivity worksheet, where its assumed price and timing are visible. It should not replace the equity footnote. Its value is to show how much an answer depends on an assumption, not to conceal the absence of a disclosed share count.

Separate earnings growth from denominator help

The historical filing reports diluted earnings per share of 4.20 and 6.36, respectively. Use the table’s numerator and denominator to decompose that change. Salesforce earnings-per-share note.

Holding the earlier denominator constant, the later numerator divided by 984 produces approximately 6.30 per share. Using the later denominator produces approximately 6.36. The denominator change contributes roughly six cents to that comparison. Most of the change comes from the numerator in this archival example.

This decomposition avoids a common exaggeration: treating all per-share earnings growth as a buyback effect. It also avoids the opposite exaggeration of ignoring the denominator. The result tells us where the arithmetic change occurred. It does not explain why net income changed; that requires examining revenue, expenses, taxes, and other items.

A hypothetical counterexample makes the distinction sharper. If earnings remain 500 million and shares decline from 100 million to 90 million, earnings per share rise from five to approximately 5.56. The reported per-share growth is about 11.1% despite no earnings growth. The company spent resources to create that denominator effect.

If those resources came from debt, interest expense can later affect the numerator. If they came from cash, the business loses the earnings or strategic options associated with that cash. A denominator bridge is therefore the beginning of capital-allocation analysis, not its end.

For a comparison across companies, place numerator growth and denominator change beside each other. It becomes easier to see whether improving per-share results came from operating performance, capital distribution, or both. A single earnings-per-share growth rate can compress those mechanisms into a number that sounds more uniform than the underlying economics.

Test a buyback price without using a current quotation

A repurchase can improve an owner’s percentage while still destroy economic value if the company pays too much. Ownership percentage and value per share are related but not equivalent. A hypothetical balance-sheet example makes this visible.

Assume a business has operating assets worth 900 million and excess cash of 100 million, for total equity value of one billion. It has 100 million shares, giving a calculated value of ten per share. Suppose it spends all 100 million cash repurchasing shares at twenty each. It retires five million shares and leaves 900 million of value across 95 million shares, or about 9.47 per share.

The remaining owner’s percentage rises because the denominator falls. Yet calculated value per share declines under the stated assumptions. The company transferred more value to departing owners than the value of the claims it retired. No real company valuation is implied; the example isolates the price mechanism.

Now suppose it repurchases at five per share. It retires twenty million shares, leaving 900 million across 80 million shares, or 11.25 per share. The same cash outlay produces a different result because the assumed repurchase price differs. The key variable is the price paid relative to the analyst’s justified value estimate.

That estimate is uncertain in practice. The example does not authorize an analyst to assign an arbitrary fair value and call every disagreement wasteful. It shows why the repurchase announcement cannot, by itself, establish success. The value judgment needs evidence about future cash generation and other claims.

This framework also clarifies compensation offsets. Buying shares solely to hold the outstanding base steady still has a price-dependent resource cost. “No net dilution” is a statement about claims after offset, not a statement that employees cost nothing or that every repurchase dollar was well allocated.

Choose one compensation convention before valuing cash flows

There are several ways to incorporate equity compensation in valuation. An analyst can model future dilution explicitly, adjust projected cash flows for an economic compensation burden, or use another coherent method. The essential requirement is consistency between the cash-flow numerator and the ownership denominator.

Consider the hypothetical 100-million-share company earning 200 million of cash before equity compensation effects. Employees receive two million additional shares annually. One model can project that issuance and allocate future equity value across a growing share base. Another can model cash spent to neutralize the issuance and subtract that spending from cash available to continuing owners. They answer related questions under different assumptions.

Using both adjustments without explaining the overlap can penalize the same issuance twice. Conversely, adding back recognized expense while holding shares permanently fixed and assuming no offset cost can omit the burden entirely. The worksheet should identify which mechanism represents compensation and where it enters.

A simplistic subtraction of recognized expense from operating cash flow can be a conservative analytical proxy, but it does not automatically reproduce the cash that would be required for an equivalent cash-compensation plan. Employees may respond differently to a changed package, and award timing matters. Label the adjustment as a model choice.

Likewise, subtracting all buyback spending as though it were a recurring operating requirement treats discretionary ownership reduction as operating cost. A business might choose to stop repurchasing; compensation issuance could continue. Split the assumed offset spending from repurchases intended to reduce the base beyond the offset.

The model need not achieve false precision. A sensitivity range with clearly stated compensation and issuance assumptions can be more informative than a single adjusted number. The discipline is to make the same economic burden appear once and to make the investor’s ownership claim agree with the modeled capital allocation.

Read award overhang as capacity and claims, not immediate dilution

The April 2025 Salesforce proxy reports approximately 39.76 million outstanding awards, excluding its employee purchase plan, and approximately 49.63 million shares available for future equity grants. It describes their combined amount as roughly 9.3% of outstanding common stock. These are historical proxy figures, and the reserve is not the same as shares already issued. Salesforce 2025 proxy, equity-plan discussion.

An outstanding award is a potential claim with specific terms. A reserve is authorization capacity for future grants. Combining them can describe the scale of possible equity-plan use, but treating the whole sum as tomorrow’s issuance ignores vesting, cancellations, exercise conditions, and management’s future decisions.

A hypothetical reserve of ten million shares illustrates the distinction. If a company grants one million annually, the reserve has a different practical meaning than if it grants five million annually. If many grants replace forfeited awards, gross grants can differ from net issuance. The reserve’s percentage alone does not establish annual dilution.

The investor can nevertheless use it to ask governance questions. How quickly is capacity consumed? What portion of awards is performance-linked? How does grant practice respond to a falling share price? Does the company seek more authorization before earlier capacity is used? These questions connect the proxy to an economic forecast without equating a permission with an accomplished event.

Award counts should also be interpreted by instrument. An option with a substantial exercise price differs from a unit that delivers a share on vesting. A headline count that treats both as identical immediate claims can mislead. The earnings denominator reflects a specific accounting calculation; it is not a complete standalone forecast of every possible future ownership outcome.

Reading the proxy and the 10-K together is useful precisely because they have different jobs. One helps describe compensation arrangements and authorization. The other reports financial results and award activity. Their dates differ, so comparisons should retain the date rather than silently merging them into one timeless snapshot.

Three objections that deserve more than a slogan

“It is noncash, so why subtract it?” Noncash recognition is relevant to liquidity. It does not establish that compensation has no economic consequence for an owner. A business can preserve cash while granting claims on future value. The analysis should identify the ownership effect or the cash needed to offset it, rather than treating the accounting classification as a conclusion.

“Employees create value, so dilution is harmless.” Employees can create substantial value. That is a reason compensation may be worthwhile, not a reason its cost should vanish from measurement. The useful comparison is the value generated relative to the total compensation package and the capital required to sustain the business. A productive employee and an expensive compensation structure can coexist.

“The share count fell, so the issue is solved.” A falling base establishes a net result for a chosen measure. The company may have paid considerable cash to achieve it. Future awards can also alter the trajectory. Evaluate the result, its cost, and the pipeline separately instead of turning the current denominator into a permanent assumption.

These objections often arise because participants are answering different questions. A chief financial officer discussing liquidity, an employee discussing incentives, and an investor discussing continuing ownership can all use the same expense line while caring about different outcomes. A good analysis makes the question explicit before comparing conclusions.

It also avoids treating equity compensation as inherently good or bad. An early business preserving scarce cash faces different tradeoffs from a mature business with substantial recurring cash generation. The test is whether the compensation and allocation policy produces an attractive outcome for the relevant owner under realistic assumptions, not whether one accounting label fits an ideological preference.

Write the investment note around the reconciled mechanism

A useful note can state that the historical period-end share base contracted, that the weighted-average diluted denominator also fell, and that repurchases consumed a material portion of the chosen cash-flow measure. It can then show the numerator-versus-denominator earnings bridge and identify what remains unknown about future issuance and execution prices.

That note is more informative than “SBC is terrible” or “the buyback eliminates dilution.” It gives another reader inputs they can challenge. They can disagree about future awards, compensation economics, or intrinsic value without first disentangling inconsistent arithmetic.

The same worksheet can be reused on another filing, but the conclusion cannot be copied. A company with rising shares despite large repurchases, a company with declining shares and no material awards, and a company issuing acquisition consideration have different mechanisms. Let the bridge determine the argument.

For the dated Salesforce example, the filing supports a historical description, not a present buy-or-sell judgment. A current investment decision would require updated filings, the current capital structure, a defensible forecast, and a price. The archival exercise has a narrower outcome: the reader can distinguish expense, claims, and cash, and avoid counting the same compensation burden twice.

Sources

• Salesforce fiscal 2025 Form 10-K filing index, filed March 5, 2025.
• Salesforce fiscal 2025 Form 10-K, financial statements, compensation notes, and earnings-per-share reconciliation.
• Salesforce 2025 proxy statement, equity awards and shares available for future grant as of March 31, 2025.