← Quantery Blog

Stock-based compensation and cash flow, explained

September 28, 2026 · 8 min readfundamentalscash-flowdilution

Stock-based compensation is pay delivered in shares or share-linked awards. Accounting records an expense, but no cash leaves the company when that expense is recognized. Under the indirect cash-flow method, the expense gets added back to net income when calculating operating cash flow. That add-back is mechanically correct. It doesn't make the compensation free.

The cost moves somewhere else. Existing owners may end up with a smaller percentage of the business as employee awards turn into shares. A company can spend cash on repurchases to offset that dilution. Or the share count can shrink despite generous stock pay if repurchases exceed issuance. Read the cash-flow add-back, the share count, and the repurchase line together.

Why does stock-based compensation boost operating cash flow?

Start with how the three financial statements fit together. The income statement measures performance under accrual accounting. The cash-flow statement starts with net income and reconciles it to cash. The balance sheet and statement of stockholders' equity record what happens to the owners' claim.

Suppose a company recognizes an illustrative stock-based-compensation expense. The income statement's compensation expense reduces net income. Yet recognizing that expense sends no payment out of the bank account. The indirect cash-flow statement therefore adds the expense back while reconciling net income to cash from operations.

net income
+ non-cash stock-based compensation
+ other non-cash and working-capital adjustments
= operating cash flow

Calling the add-back a cash-flow trick misses the accounting. Leaving the expense buried in net income would understate operating cash because no cash moved at recognition. The real mistake comes later, when a reader treats the resulting operating cash flow as though employees supplied labor for nothing.

This is the same distinction behind operating cash flow versus net income. An accrual can reduce earnings now without reducing cash now. The reconciliation explains the timing and form of payment. It doesn't judge the economics of that payment.

There can be another wrinkle. Some employee compensation is capitalized into an asset and expensed over time. That's why filings may label the cash-flow adjustment as stock-based compensation net of amounts capitalized. Read the company's label before trying to force the number to equal a stock-pay expense elsewhere in the filing.

Where does the economic cost go?

Usually into ownership, cash, or both.

When restricted stock units vest or options are exercised, the outstanding share count may rise. Each old share then represents a smaller fraction of the company. The statement of stockholders' equity shows the roll-forward. The income statement's basic and diluted weighted-average counts show how the denominator affected per-share earnings during the period. The filing cover shows a point-in-time count, which answers a different question.

That difference matters. Weighted-average diluted shares include the period's timing and potential dilutive instruments. A cover-page count captures one day's shares outstanding. A comparison between them can manufacture a dilution story because the measures describe different periods. Our article on share count and dilution walks through those definitions and the financing section in detail.

Repurchases complicate the picture again. A company may buy enough shares to absorb employee issuance, leaving the net count flat. Owners avoided dilution, but cash left the business. That cash appears under financing activities, below operating cash flow. So operating cash flow remains untouched even when the company spends heavily to prevent stock pay from expanding the denominator.

Don't automatically call every repurchase an employee-compensation bill. Companies also repurchase shares to return capital or change capital structure. The filing may not assign each repurchased share to a motive. Treat buybacks as a second observable consequence, then inspect management's authorization and the stockholders' equity roll-forward.

Snowflake shows why one cash number isn't enough

Snowflake's fiscal-year release gives a clean public example. For the year ended January 31, 2026, the company reported revenue of $4.684 billion, a GAAP net loss of $1.329 billion, and $1.222 billion of operating cash flow. The condensed cash-flow statement added back $1.600 billion of stock-based compensation, net of capitalized amounts, and showed $874 million spent repurchasing common stock.

Those figures don't support the lazy conclusion that cash flow was fake. Operating cash flow measures cash from operations under the accounting presentation, and Snowflake generated it. They do show why reported operating cash flow alone doesn't settle whether the business paid for labor in a way favorable to continuing owners.

The same release reports GAAP weighted-average basic and diluted shares of 337.493 million for fiscal 2026, compared with 332.707 million for fiscal 2025. That's an increase of about 1.4 percent, even with the repurchases. The period averages don't reveal every issuance and retirement during the year, but they give the owner-level question a direction.

Snowflake also presented non-GAAP free cash flow of $1.120 billion and defined it as operating cash flow reduced by purchases of property and equipment and capitalized software-development costs. Free cash flow therefore charges capital investment. It still starts from operating cash flow, where stock compensation has been added back. Subtracting stock-based compensation as a research adjustment produces a much harsher lens, one the company doesn't call free cash flow.

That adjusted number isn't GAAP and shouldn't be passed off as a universal definition. It asks a narrower question: would current operating cash generation remain after charging stock compensation as though it were a cash wage? Using Snowflake's fiscal 2026 figures, $1.222 billion of operating cash flow minus $1.600 billion of stock-based compensation is negative by about $378 million. It isn't a claim that the company owed an extra cash payment of that amount.

Stock awards have option value, vesting conditions, tax effects, and a grant-date accounting expense that needn't match the value employees ultimately receive. A dollar-for-dollar subtraction is deliberately severe. Use it as a stress test, then inspect what happened to shares and repurchase cash.

How should a screen test stock-based compensation?

Use three separate questions.

Did the operations produce cash? Operating cash-flow margin compares operating cash flow with revenue. It keeps the test scaled across companies.

Does cash generation survive a stock-pay charge? Subtract trailing stock-based compensation from trailing operating cash flow, then divide by revenue. This is the conservative research adjustment above.

Did each share keep its claim? Compare diluted shares with the same fiscal quarter a year earlier. If the count rises materially, the ownership cost is reaching the denominator. If it doesn't, check whether repurchases consumed cash to hold it down.

The shape below is illustrative. The bundled templates and Quantery's DSL reference are the reference for exact fields. The thresholds are proposed parameters, not facts about what every business should pass.

# "Cash generation after charging stock pay" (illustrative)
params:
  ocf_margin_floor:      0.10
  adjusted_margin_floor: 0.03
  dilution_ceiling:      0.02

features:
  # Explicit lags make the sums null if any quarter is missing.
  revenue_ttm: lag(revenue, 0) + lag(revenue, 1) + lag(revenue, 2) + lag(revenue, 3)
  ocf_ttm: lag(operating_cash_flow, 0) + lag(operating_cash_flow, 1) + lag(operating_cash_flow, 2) + lag(operating_cash_flow, 3)
  sbc_ttm: lag(stock_based_comp, 0) + lag(stock_based_comp, 1) + lag(stock_based_comp, 2) + lag(stock_based_comp, 3)

  ocf_margin: if(revenue_ttm > 0, ocf_ttm / revenue_ttm, null)
  ocf_after_sbc: ocf_ttm - sbc_ttm
  adjusted_margin: if(revenue_ttm > 0, ocf_after_sbc / revenue_ttm, null)

  shares_now:   newest(shares_diluted)
  shares_prior: lag(shares_diluted, 4)
  share_growth: if(shares_prior > 0, shares_now / shares_prior - 1, null)

criteria:
  reported_cash:
    rules:
      - { when: "is_null(ocf_margin)", score: 0, flag: no_ocf_margin }
      - { when: "ocf_margin >= $ocf_margin_floor", score: 2 }
      - { when: "ocf_margin > 0", score: 1 }
      - { else: 0 }
  sbc_charge:
    rules:
      - { when: "is_null(adjusted_margin)", score: 0, flag: no_adjusted_margin }
      - { when: "adjusted_margin >= $adjusted_margin_floor", score: 2 }
      - { when: "adjusted_margin >= 0", score: 1 }
      - { else: 0 }
  dilution:
    rules:
      - { when: "is_null(share_growth)", score: 0, flag: no_share_history }
      - { when: "share_growth <= 0", score: 2 }
      - { when: "share_growth <= $dilution_ceiling", score: 1 }
      - { else: 0 }

gate:
  mode: strict

Keep the null guards. Quantery's ttm() sums whichever of the newest four observations are present, so the example instead adds four explicit lags. Arithmetic with a missing lag returns null, which prevents an incomplete revenue, operating-cash-flow, or stock-pay window from earning a score. lag(shares_diluted, 4) assumes contiguous quarterly history, so require enough periods and audit gaps before treating it as a true year-over-year comparison.

Don't hide all three observations inside one composite feature. A company with excellent operating cash flow and modest dilution is different from one with weak cash flow and aggressive repurchases, even if some weighted total gives them the same score.

What would a backtest establish?

A useful hypothesis is that companies whose operating cash flow remains positive after the conservative stock-pay charge, while keeping share growth contained, produce different subsequent outcomes from companies that pass only the reported cash-flow test.

Run at least two versions. The control uses operating cash-flow margin and ordinary quality rules. The treatment adds the stock-pay charge and share-growth criterion. Keep the universe, rebalance schedule, filing lag, and valuation rules fixed. Otherwise you won't know which change produced the difference. Every result is hypothetical and excludes costs.

Then perturb the thresholds. Separate young software companies from mature businesses, because stock pay plays a different financing role across those groups. Inspect repurchase cash for the passing names. Compare results per share as well as at the company level. A backtest can't make the accounting choice for you, but it can show whether your choice mattered historically.

The useful conclusion may be that the severe subtraction rejects too many businesses whose dilution stayed contained. Fine. Loosen it, replace the hard gate with a score, or test share growth without the cash adjustment. Stock-based compensation forces a judgment call. Put that call in the rules where you can see it, change it, and test what happens.

Want to try this on your own rules? Quantery is free for 14 days: the full app, no card required.

← All articles