A DAO grants program pays teams outside the DAO to build things the DAO wants, out of the DAO's treasury, under rules set before any money moves. Every program settles three questions: who chooses the grantees, when the money leaves the treasury, and how anyone finds out afterwards whether the grants worked.
This page follows three programs through those questions. In Arbitrum's Short-Term Incentive Program, token holders voted on every application. Compound handed a fixed budget to one named person per area of work. Optimism elects a small council each season to choose grantees, and a second council to pay them. Retroactive rounds, which pay for work after it is done rather than before, are covered on retroactive public goods funding.
Three arrangements recur: token holders vote on each application; token holders elect a council and leave the choice to it; or the DAO gives one named person per domain a budget to spend. A direct vote puts every grant to the holders; a council or an allocator gives each application time and expertise that thousands of voters cannot.
The Short-Term Incentive Program (STIP), put to Arbitrum DAO in September 2023, set aside up to 50m ARB for protocols already live on Arbitrum to spend on incentives for their own users. Each protocol posted an application on the forum, a delegate sponsored it, and ARB holders voted on it in a Snapshot poll. An application passed with a majority in favour and more than 71.51m ARB taking part in the vote. If approved requests exceeded the budget, the proposal ranked them by votes in favour and then by the time each was posted.
The budget ran out in the first of the two planned rounds. StableLab, the firm paid to manage the process, posted during the vote that 50m ARB had been exceeded and that there would be no second round. L2BEAT's governance team, an Arbitrum delegate, wrote afterwards that requests to review applications came in such volume that it gave feedback only where it could give the same to every applicant, and assessed the finalists against a rubric of its own.
Questbook, which builds grants software, proposed in June 2022 that Compound replace a single grants committee with domain allocators: one community member per area of work, each with a budget to spend at their own discretion. The proposal named two failings of a committee: no member has expertise across every domain, and one group can only process so many applications before it burns out. Each grant was to be paid from a multisig held by the allocator and one person from the Compound Foundation, whose role was to confirm the grant fitted the domain rather than to second-guess it. Every quarter the foundation and the community could discontinue a domain, raise its budget or replace its allocator. The renewal proposal records that the program went live in January 2023 with $800,000 across four domains. The software side is covered on Questbook.
The Optimism Token House elects a Grants Council at the start of each season. Under the Season 7 charter, posted in December 2024, the council had a lead, an operations member and 7 reviewers in two subcommittees: 3 generalist reviewers who checked eligibility and alignment and guided applicants, and 4 final reviewers with DeFi experience who judged each project's potential to grow total value locked (TVL), the season's target. Members disclose conflicts of interest and recuse themselves or publish a written rationale. How the council's published scores relate to its decisions is covered on scoring and review gates.
A grant leaves the treasury in one of three ways: in full upfront, in tranches as the grantee completes agreed milestones, or as a stream paid every few weeks. Milestones and streams both keep unpaid money in the DAO's hands, so a grantee who stops delivering stops being paid.
Arbitrum streamed. The STIP sent its 50m ARB to a 5-of-9 multisig whose signers acted as grant managers for the DAO, and the DAO could claw the funds back if the signers broke the agreement. The multisig streamed ARB to each grantee every second week through Hedgey, a token-streaming tool, and could halt a stream for any use of funds not described in the application or for missing data reports, publishing its reasons on the forum. The proposal said the multisig was not there to judge the design of anyone's incentives, only "to stop bad actors, not punish bad designs". Grantees had to keep the ARB unconverted and pass KYC with the Arbitrum Foundation. Unspent funds went back: OpenBlock's review counts about 736,000 ARB returned when the program closed.
Optimism separates choosing from paying. Every grant follows the Collective's grant policies, which set milestones, bar selling the tokens and impose lock-ups. A separate Milestones and Metrics Council, a lead and 3 reviewers under its Season 7 draft charter, decides whether each grantee has met its milestones and releases the next payment from its own multisig. It can halt payments if it finds fraud or a grantee deviates significantly from its agreed milestones.
Compound paid on milestones in COMP, and the token's price moved between the decision and the payment. The renewal proposal reports that some grantees received less than they had been promised because of COMP's price and payment delays, and proposed paying later grants in stablecoins.
Mostly from metrics the grantees report themselves, from analyses by firms the DAO pays to check them, and from the program's own retrospective against the targets it set at the start.
Arbitrum required each STIP grantee to publish a dashboard of daily TVL, transactions, volume, unique addresses and fees covering 30 days before, during and after its incentives. OpenBlock's efficacy analysis, covering the 30 protocols funded, found that their combined TVL rose 118.2% between 3 November 2023 and 29 March 2024, against 35.6% for Arbitrum protocols outside the program. The same report notes that the growth of Pendle, which led all grantees on TVL growth per ARB claimed, was also driven by its recent tokenization of points. Chaos Labs, reviewing three grantees for the DAO's research collective, found little evidence that the incentives had caused bad debt, cascading liquidations or other adverse outcomes. It attributed that partly to the program starting in a bear market, and recommended that future applications declare their risks.
The Optimism Grants Council's Season 7 retrospective reports that it met its internal targets: reviews under 2 weeks, every application reviewed, 3 reviewers at intake and 4 at the final stage. About 10m OP had been granted by then; grantees reported using 356,000 OP and $480m of TVL attributed to the grants, $400m of it from one project, Spark. The council cautioned that the figures were early. It also listed what went wrong: first feedback took 3 weeks to reach applicants, and some projects applied to more than one Optimism funding stream, risking duplicate OP for similar goals. It credited a cut in its own size from 12 members to 7 with better coordination.
Compound's two-quarter report shows the gap milestone payment opens between promising and paying. Allocators accepted about 40% of more than 100 proposals and approved $670,000 of grants, but had disbursed $252,000, because grantees are paid as they complete milestones. The report puts the milestone completion rate at about 57% for funded proposals and about 27% across all accepted proposals.
A caper has no separate grants council or allocator. Its treasury spends only through a proposal its holders vote on, so every grant is a holders' decision, made one proposal at a time. A payout proposal that passes pays the named recipient directly from that caper's treasury. Members who disagree with how the treasury is spent can leave with their share. Paying someone from a caper walks through a payout, and proposals covers how a vote runs.