Our Rule of 40 explainer lists the metric's blind spots, and one of them is a sentence we have never tested:
Stock-based compensation is invisible in the EBITDA version. A company paying 30% of revenue in SBC looks profitable on EBITDA but is diluting shareholders heavily. The FCF version partially captures this, but not perfectly.
This site uses the FCF version. So the question that sentence leaves open is the only one worth asking: how imperfectly?
Here are the exact four companies the same explainer names in "Where it does work", each at its latest reported quarter, plus Unity as a fifth row for a reason that becomes obvious.
| period end | Rule of 40 | free cash flow | SBC | SBC / revenue | SBC / FCF | less SBC | |
|---|---|---|---|---|---|---|---|
| Datadog | 2026-03-31 | 64.2 | $323.2M | $196.8M | 19.55% | 60.9% | 44.65 |
| ServiceNow | 2026-03-31 | 62.6 | $1,529.0M | $547.0M | 14.51% | 35.8% | 48.09 |
| CrowdStrike | 2026-04-30 | 59.5 | $468.5M | $297.7M | 21.49% | 63.5% | 38.01 |
| Palo Alto Networks | 2026-04-30 | 57.4 | $788.0M | $643.0M | 21.42% | 81.6% | 35.98 |
| Unity | 2026-03-31 | 29.8 | $66.5M | $77.2M | 15.19% | 116.1% | 14.61 |
Two of the four exemplars fall below 40. The SBC share of revenue is carried to two decimals so the last column is exactly the first column minus it — rounded to one decimal, Datadog's row would read 64.2 − 19.6 and give 44.6, which is not the number beside it.
Between 35.8% and 81.6% of the free cash flow that makes up the profitability half of these scores is matched, in the same quarter, by stock issued to employees. Unity inverts it entirely: its stock compensation is larger than its free cash flow.
One caveat on the window, since this site has just finished documenting how much damage a single quarter can do. Every row above is one reported quarter, and the free-cash-flow denominator is the volatile term — Netflix's quarterly Rule of 40 fell 32 points while its trailing-twelve-month score moved three, purely because its largest cost is weighted to the first half of the year. Stock compensation is the steadier of the two figures here; it is set by grant schedules rather than by payment timing. So SBC / revenue travels better across quarters than SBC / FCF, and a company whose cash happened to land in a weak quarter will show a worse ratio than its year does. Read the first of those as the durable figure and the second as a snapshot.
The last column is one adjustment, not a correction. Whether stock compensation should be subtracted in full is a live disagreement — it is precisely the argument companies make in their own non-GAAP reconciliations, in the opposite direction. We are not proposing a new formula or restating any score. Read the two columns together; that is the whole point.
Why the FCF version misses it — twice
Stock-based compensation is a non-cash expense, so it is added back in operating cash flow. Free cash flow is operating cash flow minus capital expenditure, so the addback flows straight through to the profitability half of the score. A company that settles a fifth of its payroll in shares reports free cash flow roughly a fifth of revenue higher than one paying the same people in cash — and the Rule of 40 rewards it for the difference.
The contrast with the other way a company can pay for its future is stark, and it landed on this site the same day. Capital expenditure is subtracted from free cash flow in full and immediately, which is why Meta fell below 40 for the first time in three years on capex running at 49.5% of revenue — a domain failure, where the metric is describing the company correctly and the company is leaving the range the metric was built for. Buy the asset with cash and the score charges you for all of it in the quarter you buy it. Pay the engineers who build the asset in stock and the score charges you for none of it, ever. Both are the company spending something real to grow; only one is visible in the number.
The cash spent mopping it up sits below the line as well. Share repurchases are a financing activity. They never touch operating cash flow, so they never reduce free cash flow, and the metric cannot see them either.
That second half matters for a question a careful reader will ask: isn't subtracting SBC double-counting, given these companies also buy stock back? No — and specifically because the buyback was never inside free cash flow to begin with. The metric adds the stock back and then cannot see the cash spent retiring it. Those are two separate omissions pointing the same way.
The three companies handle the second one very differently, which is what makes this about corporate choice rather than a single accusation:
- Datadog files no repurchase concept at all. Not a zero — the concept is absent from its filings, so there is no buyback programme offsetting the issuance. The dilution runs straight through to shareholders.
- CrowdStrike repurchased $175.6M in the same quarter it expensed $297.7M of stock compensation — roughly three fifths of it.
- Palo Alto Networks repurchased nothing in FY2025 and $1,000.0M in the nine months to 2026-04-30, in a stretch whose latest quarter also carried a GAAP loss. That quarter is its own story and is covered separately; the $643M of stock compensation in this table is a large part of its mechanism.
The column that does not do what you would expect
The obvious next step is to treat SBC as a proxy for dilution. It is not one, and the data says so plainly. Diluted share counts, comparing each company's latest quarter against twelve quarters earlier:
| SBC / revenue | diluted shares, 3 years | |
|---|---|---|
| Datadog | 19.55% | +5.5% |
| CrowdStrike | 21.49% | +11.6% |
| Unity | 15.19% | +47.5% |
Datadog pays the second-largest share of its revenue in stock and diluted the least. Unity pays the smallest share of the three and diluted nearly nine times as much. The difference is share price and scale, not generosity: the same dollar of stock compensation buys far fewer shares at a high price, and a large share base absorbs more issuance before the percentage moves.
So the two measures answer different questions. SBC as a share of revenue tells you how much of the profitability half of the score is being paid in stock. Share count tells you what that cost shareholders. Neither substitutes for the other.
ServiceNow and Palo Alto are absent from that table deliberately, and the reason is a trap worth naming. Both ran stock splits inside the three-year window — ServiceNow's series jumps roughly fivefold between the quarters ended 2025-09-30 and 2026-03-31, Palo Alto's doubles between 2024-10-31 and 2025-01-31. A naive comparison of raw share counts reports Palo Alto as having diluted more than 700% and ServiceNow more than 400%. Both figures are artifacts of the split, not dilution, and any table printing them would be nonsense. Three rows with the reason for the exclusions stated beats five rows where two are absurd.
What to do with this
The site is not changing its formula, and this piece is not asking it to. The useful habit is smaller and available to anyone: when you look at a Rule of 40 score, read the free-cash-flow margin next to stock compensation as a share of revenue. Both are public, both come from the same filing, and together they tell you how much of the profitability half was paid for in cash and how much in shares.
On the explainer's own four exemplars, that ratio runs from a third to four fifths. The metric is still doing its job on these companies — growth and cash generation really do trade off against each other here, which is why the explainer chose them. It is just doing it on a cash figure that counts a large share of payroll as free.
That habit is the same one Alphabet's quarter argues for from the opposite direction: its score moved six hundredths of a point in a year while both halves moved more than ten, so the composition carried the entire story and the headline number carried none of it. That is a composition failure — a sum cannot show you a trade — and this is an omission one, where a real cost never enters either half. They arrive at the same instruction from different ends. The score is the least informative thing on the page; what it is made of, and what it left out, are the rest of it.
Stock-based compensation is ShareBasedCompensation from SEC XBRL for each company: Datadog CIK 0001561550 ($196.8M), ServiceNow CIK 0001373715 ($547.0M), CrowdStrike CIK 0001535527 ($297.7M), Unity CIK 0001810806 ($77.2M) — each tagged directly — and Palo Alto Networks CIK 0001327567 ($643.0M), derived as the nine-month $1,314.0M less the six-month $671.0M. Buybacks are PaymentsForRepurchaseOfCommonStock at the same CIKs; Datadog returns no such concept. Diluted share counts are WeightedAverageNumberOfDilutedSharesOutstanding, quarterly points only. Published Rule of 40 scores and FCF margins are the values this site renders, computed as revenue growth plus free-cash-flow margin on the house definition in our explainer; Datadog, ServiceNow, CrowdStrike, Unity and Palo Alto revenue and free cash flow were each reconciled against XBRL at the CIKs above. Note that CrowdStrike files revenue under RevenueFromContractWithCustomerIncludingAssessedTax and capitalises software development alongside property and equipment, so both are needed to reproduce its free cash flow. Free cash flow is shown so every ratio in a row can be checked from that row; Unity's is the $66.5M the filing derivation gives ($71.3M operating cash flow less $4.8M capex), where this site's stored series rounds it to $66.0M — which is why that ratio is 116.1% here and 117% if taken from the rounded figure. As the explainer notes, two of these quarters end in March and two in April, so they are close to like-for-like rather than identical periods.