How to Treat Stock-Based Compensation in Valuation Math

Fifty-nine cents. That's how much of every dollar of free cash flow Workday reported in its most recent fiscal year was matched by a non-cash charge for stock paid to employees. According to Workday's 10-K for fiscal 2026 (the year ended January 2026), operating cash flow was $2,939 million and capital expenditures were $162 million, which leaves $2,777 million of free cash flow. The same cash flow statement adds back $1,626 million of share-based compensation to get there.

That add-back is legitimate accounting. No cash left the building when those shares were granted, so the cash flow statement is right to put the expense back. But a valuation isn't a cash flow statement. It's an estimate of what each existing share is worth, and a share is a claim on a business whose future cash is being promised, in part, to people who don't own it yet. Whether you count that promise as a cost or as dilution, you have to count it once. Skip it and the model comes out wrong by a wide margin.

There are two defensible ways to do it. I'll run both on the same Workday numbers, with the same discount rate and growth rate, and show where they agree, where they don't, and what the gap between them is telling you. Claire Nakamura has already covered the accounting side in what the stock-based compensation add-back conceals; this piece is about the valuation math that comes after.

Two Treatments That Should Land in the Same Place

The first treatment says stock compensation is an operating expense like any other. Employees work, the company pays them, and the currency happens to be shares. Under this view you subtract it from free cash flow before you do anything else, and you divide by today's share count.

Treatment 1 (expense it): Owner cash flow = Operating cash flow − Capital expenditures − Stock-based compensation. Value per share = Owner cash flow ÷ Current diluted shares, capitalized at your discount rate.

The second treatment keeps the cash flow statement's add-back but refuses to pretend the share count is fixed. You value the full free cash flow, then shrink each existing share's slice of it by the rate at which new shares are being issued. Per-share cash flow grows more slowly than the business does, and the difference is the dilution.

Treatment 2 (add back, model dilution): Per-share growth = (1 + business growth) ÷ (1 + annual dilution rate) − 1. Value per share = Free cash flow per share × (1 + per-share growth) ÷ (discount rate − per-share growth).

In theory these are the same thing seen from two sides. A company that pays employees $1.6 billion in stock could instead have paid them $1.6 billion in cash and then sold $1.6 billion of new shares to raise it. Under the first framing the cost sits in the cash flow. Under the second it sits in the share count. If the dilution rate you pick in Treatment 2 truly matches the value of the grants, the per-share answers converge.

In practice they rarely do, because almost nobody checks whether their dilution assumption is consistent with the size of the grants. That's the error I want to make visible. It's also why the naive version — add back the expense and hold the share count flat — isn't a third treatment at all. It counts the employees' slice as if it belonged to you. The same distortion carries straight into FCF yield: on the Workday numbers below, free cash flow with the add-back is about 2.4 times free cash flow after stock compensation ($2,777M ÷ $1,151M), so the yield looks 2.4 times as generous at any given price.

Workday, Run Both Ways

Workday is a clean test case. It's profitable on a cash basis, its stock compensation is large relative to that cash, and it has been buying back shares heavily. Everything below comes from Workday's 10-K filings, with fiscal years ending in January. All per-share figures use weighted-average diluted shares of 268.1 million for fiscal 2026.

• Free cash flow, fiscal 2026: $2,939M − $162M = $2,777M, or $10.36 per share.
• Stock-based compensation, fiscal 2026: $1,626M, about 17.0% of revenue of $9,552M and 58.6% of free cash flow.
• Free cash flow after stock compensation: $2,777M − $1,626M = $1,151M, or $4.29 per share.

Now the assumptions, stated plainly because the answer depends on every one of them. Discount rate: 9%. Long-run growth in the business's cash flow: 5% a year. I'm using a simple growing-perpetuity model rather than a multi-stage DCF so the effect of the stock compensation treatment isn't buried under other moving parts. These are illustrative inputs, not a forecast of Workday's future.

1. Naive add-back, no dilution: $10.36 × 1.05 ÷ (0.09 − 0.05) ≈ $272 per share.
2. Treatment 2, add back with 2.5% annual dilution: per-share growth = 1.05 ÷ 1.025 − 1 ≈ 2.44%, so $10.36 × 1.0244 ÷ (0.09 − 0.0244) ≈ $162 per share.
3. Treatment 1, expense it: $4.29 × 1.05 ÷ (0.09 − 0.05) ≈ $113 per share.

Same company, same year, same discount rate, same growth rate. The estimate of value runs from $113 to $272 depending only on how one line of the cash flow statement is handled. None of those numbers is a price, and I'm not comparing any of them to where Workday's shares trade. They are three estimates of value, and the spread between them is the point.

So which dilution rate makes Treatment 2 agree with Treatment 1? Working backward, per-share growth would have to be slightly negative, about −0.2% a year, which means dilution of roughly 5.2% a year against 5% business growth. That sounds high. But check it against the grants themselves. Treatment 1 values the whole equity at about $113 × 268.1 million ≈ $30.2 billion. Annual grants of $1,626 million are about 5.4% of that figure. The two treatments converge when the dilution rate equals the annual stock grants divided by the value of the equity, which is exactly what you would expect if stock compensation is a real cost paid in shares.

Consistency check: Implied annual dilution ≈ Annual stock-based compensation ÷ Value of the equity. If your Treatment 2 dilution rate is far below this ratio, you are adding back the expense without charging for it.

That ratio is where I'd put most of the attention. A 2.5% dilution assumption looks conservative on its own. It's only when you set it against $1.6 billion of annual grants that you see it implies an equity value above $60 billion — roughly twice what the expensed cash flow supports. The model is quietly assuming its own conclusion.

What the Share Count Says, and What It Hides

The obvious objection is that Workday's share count isn't exploding, so maybe the dilution is smaller than the grants suggest. The filings do show a fairly stable count. Weighted-average basic shares were 254.8 million in fiscal 2023 and 265.1 million in fiscal 2026, an increase of about 4.0% over three years. Diluted shares actually dipped, from 269.2 million in fiscal 2025 to 268.1 million in fiscal 2026.

But look at what it cost to hold the line. Workday's repurchases were $423 million in fiscal 2024, $700 million in fiscal 2025 and $2,895 million in fiscal 2026, a total of $4,018 million. Stock compensation over the same three years was $1,416 million, $1,519 million and $1,626 million, or $4,561 million. Four billion dollars of cash went out the door, and the basic share count still rose. Much of that buyback money was absorbing grants rather than shrinking the owners' base.

This is why the share count alone is a poor dilution input. A company can make gross dilution invisible by spending cash to offset it, and then the cost has moved from the share count back into the cash flow — just under the financing section, where an owner-earnings calculation never looks. If you use Treatment 2 with net dilution near zero because buybacks offset the grants, you have to treat the offsetting buybacks as a use of the free cash flow, not as a return to owners. Our piece on adjusting per-share value for heavy buyback activity walks through the per-share side of that.

Here I'll admit the limits of my own method. I'm less sure than I'd like to be about the right dilution rate for any one company, because grants are struck at the share price on the grant date, not at your estimate of value, and the two can sit far apart for years. A company whose shares trade well above intrinsic value pays employees with fewer, pricier shares; one trading below value pays with more. The ratio in the consistency check is approximate for that reason. It still beats an unexamined assumption.

Key Takeaways

Neither treatment is a precise answer. A valuation built either way is a compass, not a GPS coordinate, and the honest output is a range wide enough to survive a change in the dilution assumption. But the two treatments should land close together, and when they don't, the gap points to the assumption doing the work. For the broader method these pieces sit inside, our guide to calculating fair value of a stock and the comparison of owner earnings and free cash flow are good next reads.

• Stock-based compensation has to be counted once: either as an expense deducted from free cash flow, or as dilution that slows per-share growth. Adding it back with a flat share count counts it zero times.
• On Workday's fiscal 2026 numbers, at a 9% discount rate and 5% growth, the treatment alone moves the estimate of value from about $113 to $272 per share.
• The two treatments agree when annual dilution roughly equals annual stock grants divided by the value of the equity. Check your dilution assumption against that ratio before trusting it.
• A stable share count can hide large gross dilution when buybacks are absorbing grants. Workday spent $4,018 million on repurchases over fiscal 2024 to 2026 while its basic share count still rose about 4%.