Nobody Converts: A Conversion Census and Tranche Gate Before the Foundation Inherits the Growth Problem

I am not litigating custody, and I am not restating the case for pricing the issuer layer that another entry in this slate already makes at length. This proposal is scoped deliberately to the other side of that question - not who should pay the DAO, but whether anybody using an ENS name ever becomes a customer of one. It is the companion to my earlier filing here on renewal ("The Most Expensive Names Are the Least Kept"), which showed that the people who pay do not stay. This one asks whether the people who stay ever start paying, and gates the Foundation's budget on the answer. THE FINDING, IN ONE COMPARISON. Primary names - the setting a person configures when they actually use an ENS name as their identity - stand at 385,562 on Ethereum mainnet against 3,093,102 on Base (https://enswhois.com/stats/primary-names, read 2026-08-08). Eight times as many people have adopted an ENS-resolvable identity on a layer where the DAO collects no registration and no renewal as on the layer that funds it. Adoption did not stall. It relocated. THE PAID BASE IS A MAINTENANCE BUSINESS. From the same dataset (https://enswhois.com/stats, read 2026-08-08): 850,456 active .eth names held by 392,495 unique owners, with 3,716,444 names expired and never re-registered, and 283,482 more expiring within ninety days. Over the trailing ninety days the protocol recorded 32,058 new registrations against 145,844 renewals. Four and a half renewals for every acquisition is not a growth funnel. It is an installed base being maintained while a much larger population uses the product for free somewhere else. WHAT THAT DOES TO REVENUE. The share of DAO revenue from premium auction surcharges - the fee on names released after somebody let them lapse - is now over 50 percent in 2026 (Thomas Clowes, "What ENS Data Suggests About Pricing, Retention, and Speculation", May 2026: https://thomasclowes.com/what-ens-data-suggests-about-pricing-retention-and-speculation/). The DAO's own governance frontend reports trailing revenue of $5.6M, a run rate of $4.6M a year, revenue down 49 percent year on year, and new wallets at 23.5K, down 70 percent (https://ens.gov.blockful.io/revenue, read 2026-08-08). More than half of what ENS earns now comes from reselling inventory its customers abandoned, and the inflow of new customers has collapsed. THE HONEST COUNTERPOINT, STATED UP FRONT. On the same dashboard the renewal rate is 42 percent and improving, up 11 points. Retention among people who already pay is getting better, not worse. That is real and it cuts against urgency. It does not touch the argument, because renewal improves the yield on a shrinking base - it cannot replace the acquisition that has stopped. But anyone reading this should know that one of the two numbers is moving in the DAO's favour. WHAT THE EXECUTABLE SAYS ABOUT CONVERSION: NOTHING. The executable hands the Foundation the trademarks, the grants programme, SPP absorption, Endowment stewardship, and representation at ICANN, IETF and W3C including pursuit of the .ens TLD (https://www.tally.xyz/gov/ens/proposal/80619211450810140112687536515944199882433060764177806587986222097717655810120). Conversion is not mentioned, and no reporting obligation attaches to it. The Foundation will be measured on process and spending while the ratio that determines whether its budget exists in three years is nobody's responsibility. Under ENSv2, where every name gets its own registry (https://ens.domains/blog/post/ensv2-architecture), delegated free issuance becomes the protocol's native shape - so this gap widens by design, not by neglect. MECHANISM 1 - THE CONVERSION CENSUS. A quarterly published figure, per integrator cohort: of the wallets holding a free ENS-resolvable subname at the start of a period, what share held a paid second-level .eth name at the end of it. This measures people, not issuers, and it is the number no participant in this debate can currently state. Method and code published with the first release so anyone can reproduce or dispute it. MECHANISM 2 - THE TRANCHE GATE. The DAO sets a conversion floor before each period begins, expressed relative to the prior cohort rather than as an absolute, so a market-wide downturn cannot trip it. Miss the floor in two consecutive periods and the Foundation's next operating tranche is gated pending a published remediation plan. Not a clawback, not a veto - a pause with a stated cure. This is the reverse-onus structure the vault-side proposals apply to spending, applied to growth, and it composes with them rather than competing. MECHANISM 3 - PRE-REGISTERED MEASUREMENT. Definitions, denominators, observation window and data source are committed as a hash before the period they measure. Revisions are allowed but must be filed as logged deviations before the report is published; an unlogged deviation is itself a gate failure. Without this, whoever reports the number can choose it after seeing the result, and the gate becomes theatre. MECHANISM 4 - THE NO-TOLL RULE. Nothing in this proposal permits the Foundation to improve the conversion ratio by degrading the free tier. End users never pay for a free subname they already hold, and gating may not be satisfied by making free issuance worse. If the only way to hit the floor is to tax people who are currently served for nothing, the correct answer is to miss the floor and file the remediation plan. BUDGET: $9,000, phased, specification rather than implementation. $5,000 to build and publish the conversion census pipeline and its first release. $2,500 for the measurement protocol - definitions, denominators, hashing scheme, deviation register. $1,500 to draft the remediation-plan template and gate language in executable-ready form. Minimum viable version, the method plus a single first census: $2,500. Each phase stands alone and can be funded without the others. WHAT WOULD CHANGE MY MIND. If the first census shows free subname holders already acquire paid .eth names at a materially higher rate than the general wallet population, then the free layer is already functioning as the funnel, no gate is warranted, and the correct response is to fund more free issuance rather than police it. I would accept that result and say so. The census is worth building either way, because the DAO is presently deciding the shape of a nine-figure endowment without knowing whether its product converts. WEAKNESSES I WILL STATE BEFORE SOMEBODY ELSE DOES. First, attribution is imperfect: linking a paid registration to a wallet that previously held a free subname is straightforward on-chain but blind to users who move between wallets, so the census will understate conversion and should publish its error bound rather than a single figure. Second, offchain names resolved through a CCIP gateway the DAO does not operate are not fully observable, so cohort sizes for those integrators depend partly on self-reporting and will be incomplete in the first release. Third, as noted above, renewal is already improving, so a badly specified floor could penalise the Foundation for a trend that is correcting without it - which is why the floor must be cohort-relative and why two consecutive misses, not one, are required to gate. Fourth, privacy: cohort analysis of wallet behaviour is aggregate here by design, and any release granular enough to identify individuals should be withheld. All figures in this proposal were read on 2026-08-08 and each carries its source above. Where a source could not be verified directly, I have said so rather than rounding it into the argument.