Fee yield, revenue yield and holder-revenue yield are not income
These three figures invert the valuation ratios into percentages that look like yields, and none of them describes cash arriving in anyone's account.
Fee yield, revenue yield and holder-revenue yield are the valuation ratios turned upside down and expressed as percentages. A ratio of 50 becomes a yield of 2 percent. The inversion is arithmetic and harmless; the word yield is neither, because in every other corner of finance it denotes cash a holder receives, and here it generally does not.
The three are published as fee yield, revenue yield and holder revenue yield.
How each is constructed
| Measure | Numerator | Denominator | Does a holder receive anything |
|---|---|---|---|
| Fee yield | Annualized total fees paid by users | Market cap | No; most is paid to the supply side |
| Revenue yield | Annualized protocol revenue | Market cap | Only where a distribution mechanism exists |
| Holder revenue yield | Annualized value accruing to holders | Market cap | Sometimes, and often as a burn rather than a payment |
| Staking yield | New issuance plus fees paid to stakers | Staked balance | Yes, in tokens, funded largely by issuance |
The first three describe economics at the level of the system. The fourth describes a payment to a participant performing a service, computed over a different denominator entirely. Mixing them in one table is a frequent source of confusion and is done here only to make the difference explicit.
Why the first two are not income in any sense
Protocol fees are paid by users to whoever provided the resource. On a proof-of-stake network that is validators. On an automated market maker, it is liquidity providers. A holder who does none of those things receives nothing from a fee yield, no matter how large the figure is. What the number describes is the fee intensity of a network relative to its valuation, which is a real property and not a return.
Protocol revenue narrows this to the retained portion, and retained is not the same as distributed. Revenue accumulating in a protocol treasury is controlled by governance and may be spent on grants, audits, security, market making or salaries. A treasury balance is not a claim; holders generally have no mechanism to compel a distribution from it, and in most systems no mechanism to prevent a spend from it either.
The holder-revenue case, stated carefully
Holder revenue is the portion of value whose effect lands on holders specifically. Where this takes the form of a token burn, the mechanism deserves precision. No payment occurs. Units are destroyed, so each remaining unit represents a larger share of a network of unchanged size. If nothing else changes, that is economically similar to a pro-rata buyback of shares, and the qualifier carries the weight, because something else almost always changes: new units are being issued at the same time.
The net effect is captured by real inflation, which subtracts burns from issuance. A network with a visible burn and a larger issuance rate is expanding supply on balance, whatever its holder-revenue yield reads. Reading a burn-based yield without the issuance beside it counts one half of a subtraction and reports the result as a full answer.
Where holder revenue is a genuine distribution, paid to stakers or lockers of the token, three further conditions apply. It is usually paid in the protocol's own token or in an asset carrying its own risk. It is contingent on continued governance approval and can be reduced or removed by the same process that created it. And it typically requires an action, such as staking or locking, that carries smart contract risk and a withdrawal delay during which the position cannot be exited.
Staking yield, which is a different object
Staking yield is compensation for performing validation work. Its main component is new issuance, which means the payment is funded by diluting everyone, including the recipient. A staker earning at roughly the network's issuance rate is approximately holding position rather than gaining; a non-staker is being diluted. This is why nominal staking yield and real staking yield, which nets out issuance, are shown separately rather than as one figure.
The term real yield came into use in response to exactly this: distinguishing rewards funded by fees users actually paid from rewards funded by printing new units. The distinction is mechanical rather than rhetorical, and it is visible in the split between supply-side revenue and issuance on the fee pages.
Staking also carries obligations that a percentage does not convey. Slashing penalizes faults and can remove principal. Unbonding periods lock capital for a defined window regardless of what happens in the market during it. Delegating to an operator introduces counterparty risk that the nominal rate does not price. The staking pages set out these parameters per network.
Comparing any of these to a bond is a category error
A bond coupon is a contractual obligation of an identified issuer, enforceable in court, paid in a currency the holder did not have to acquire in order to hold the instrument. None of the four measures above shares a single one of those properties. There is no issuer, no obligation, no enforcement, and the unit of account is the asset itself, so a percentage stated in tokens says nothing about purchasing power in any other unit.
The measures remain useful for what they are: comparable, scale-free descriptions of how much fee activity, retained revenue or holder-directed value a network generates relative to what the market values it at. That is a legitimate object of study, and expressing it as a percentage makes assets of very different sizes comparable in a way the raw ratios do not. It is not a rate of return, and this site does not present it as one.
The fees and staking pages carry the underlying series, and the metric catalog states the exact numerator and window for each of the four.