Earlier today I had a board meeting for the nonprofit I founded, ALTER—and it was a good one. Most of the board showed up. People had looked at the materials. I presented our recent update, funding, budgets, and our plans for the coming year. There were thoughtful questions: what’s the path to impact for working on standards, some skepticism about collaborating with non-aligned organizations, and a few suggestions for adjacent things we might consider. No one was asleep. No one was hostile. No one was performatively disengaged.

But when the meeting ended, no strategic direction had been set. The board didn’t tell us what to prioritize next year. They reviewed the presented plans, and the priorities I suggested, but didn’t weigh tradeoffs between mission drift and focus. They didn’t say “this seems off” or “this is the wrong hill.” What they mostly did was react: to updates, to explanations, to decisions that had already been made elsewhere. Their feedback was real, but incremental—more like local gradient information than a global objective.

I’m on a few small organization’s boards, and at ALTER I have a board I am responsible to, and I don’t think what happened is particularly unusual. This isn’t a complaint. It’s an observation. The strategic direction is mostly coming from me, not from them. And so, even though I’ve read the advice, and know that this is all a bit weird, I still had a question I didn’t really know how to answer: is that how this is supposed to work?

So I did what any proactive leader should do - I sat down with a language model and tried to work out an answer that I could endorse.

Who sets priorities?

There’s an obvious alternative candidate for who sets priorities; funders. But that’s a partial story at best. Our funding is based on what I apply for, and at most comes earmarked by broad project areas or themes: this bucket of money is for AI policy, that bucket is for biosecurity, and another contract is specific to Vanessa’s technical AI safety work. That constrains the feasible set of actions, but which grants we receive depends on where I choose to apply, and for what. And once the funding arrives, it rarely determines which concrete projects within those areas we pursue, or how aggressively, or in what order. Funders often have views about what’s promising, and in some cases we get valuable feedback, but they rarely issue anything like a binding strategic directive.

In practice, as Holden explained, agenda-setting happens somewhere fuzzier. Ideas originate with me. Some come from conversations with funders, some from collaborators, some from noticing gaps or opportunities that feel salient at the time. I propose directions. Funders respond with interest or hesitation. The board sanity-checks the result after the fact. Everyone has influence, but no one quite owns the strategy.

Part of what makes this confusing is that “setting the agenda” isn’t a single thing. It’s a bundle of different functions that we often collapse into one phrase. There’s idea generation: noticing what might be worth doing at all. There’s prioritization: deciding which of several plausible directions to pursue, and which to drop. There’s constraint-setting: money, scope, risk tolerance, reputational boundaries. And there’s accountability: being on the hook, in some meaningful way, for whether the chosen path actually works.

In our current setup, those functions are split across actors, but not very deliberately. I generate most of the ideas and do much of the prioritization. Funders shape the constraints, but indirectly. The board provides feedback and oversight, but mostly downstream. The result is that I, as the founder, have a lot of control—but not a clean mandate, not an explicit delegation. It’s control by default, emerging from the fact that I’m the one synthesizing inputs and turning them into concrete actions. And Holden’s ideas are good ones - especially keeping it simple for small orgs - but they also don’t quite answer my questions.

“Who” depends on “What”

One reason this question is hard to answer is that we often talk about nonprofit governance as if there were a single right model, or at least a narrow range of respectable ones. But that only makes sense if nonprofits are all trying to solve roughly the same kind of problem. They aren’t.

Different kinds of nonprofits exist to do very different things. Some are trying to discover what works in an uncertain domain. Some are trying to faithfully execute a known program at scale. Some exist to carry out the priorities of donors. Others exist to respond to the needs of a particular community. Once you take that seriously, it becomes much less plausible that agenda-setting should look the same in all cases.

Consider founder-driven research nonprofits - like ALTER. These tend to operate in domains with high epistemic uncertainty, where the main bottleneck isn’t execution capacity but judgment: what questions are even worth asking, what approaches are promising, which bets are worth placing. In those settings, agenda-setting lives with the founder, not because founders are uniquely insightful or virtuous, but because they are the ones doing the synthesis. Boards in these organizations tend to play a constraining role—sanity-checking, preventing obvious failure modes, occasionally saying “no”—and funders bound the scope of exploration. But neither is well-positioned to generate strategy from scratch.

Contrast that with donor-program nonprofits, which exist primarily to implement a set of priorities defined upstream. In these cases, agenda-setting is intentionally funder-heavy. The organization’s job is not to decide what matters, but to execute well on what has already been decided. Staff optimize for delivery; boards focus on compliance, effectiveness, and risk. A board that tried to “set strategy” independently here would likely be confused about its role, and possibly disruptive.

Then there are program-delivery or service nonprofits, where the mission is stable and the risks are operational or reputational rather than exploratory. Hospitals, humanitarian NGOs, and large education providers fall into this category. Here, boards and executive leadership play a much stronger role in setting priorities, precisely because the cost of drift or failure is high. Funders are often deliberately limited in how much agenda control they have, to preserve mission integrity and continuity.

Finally, there are community-serving nonprofits—local community centers, athletic or cultural organizations—where legitimacy comes from responsiveness to the people served. In these cases, agenda-setting flows primarily from the public: through usage patterns, feedback, and local norms. Staff translate those signals into programs; boards act as stewards of trust rather than authors of vision. A founder or donor trying to impose priorities top-down would undermine the very thing the organization exists to provide.

Across these cases, the pattern is not that one group always ought to set priorities, but that agenda-setting tends to live with whoever is best positioned to solve the organization’s core problem. When the problem is discovery, it gravitates toward founders. When it’s execution on behalf of donors, it sits with funders. When it’s risk management and continuity, boards matter more. When it’s legitimacy and responsiveness, the served community dominates.

The trouble starts when an organization inherits a governance model from the wrong reference class—or when it never quite decides what problem it is trying to solve, and so never chooses who should really be in charge of deciding what comes next.

Ambiguity isn’t the problem. Unowned outcomes are.

At this point, it’s tempting to say that the problem is ambiguity—that if only we were clearer about who sets priorities, things would be better. I don’t think that’s quite right.

Many healthy organizations are ambiguous about influence. Ideas come from multiple places. Constraints come from somewhere else. Oversight is distributed. This can work surprisingly well. In fact, it often works better than rigid hierarchies, especially in domains where information is partial and the future is hard to see.

What seems more dangerous is not shared influence, but diffuse accountability. You need RACI, or something similar; someone Responsible, someone Accountable, and others Consulted or Informed. Depending on your mission and structure, the distribution of the roles will differ.

In the cases that go wrong—not just in nonprofits, but in institutions more broadly, often this is what failed. Someone could have intervened, someone who had input, but no one who clearly owned the outcome. Decisions accrete. Directions drift. Everyone can explain why the thing that happened was reasonable at the time, given the signals they saw. And yet, somehow, no one was really on the hook for whether it worked.

Boards, in particular, seem vulnerable to this pattern. Are they supposed to offer feedback? Are they meant to approve plans? Should they own consequences if things go badly? It’s often unclear. This isn’t because board members are careless or disengaged, but because the structure makes it easy to participate without anyone knowing who was responsible for what.

A board that only reacts can still feel involved. A founder who synthesizes inputs can still feel constrained. Funders who set boundaries can still feel hands-off. From the inside, everything feels collaborative. From the outside—or in hindsight—it can look like no one was steering.

Questions I don’t yet have answers to

I don’t have a neat resolution to this. I’m not trying to argue that boards should always set strategy, or that founders should relinquish it, or that funders should be more explicit. In fact, I’m confident that any blanket recommendation here would be wrong.

What seems useful is to notice questions that could go unasked:

• Who is actually accountable for the success or failure of this organization’s strategy?
• If priorities drift, who is expected to notice—and who is empowered to fix it?
• When does founder-led agenda-setting stop being a feature and start being a risk?
• When should boards move from offering gradient feedback to owning hard tradeoffs?
• What signals should trigger a deliberate shift in who sets priorities?

Most nonprofits answer these questions implicitly, through habit and inertia. Some of those answers are fine. Others persist long past the conditions that made them sensible. I’m hoping we are the former, but want to be more deliberate about this.

Because I’ve realized that the problem isn’t that we don’t have a clean story about who sets the agenda. The problem is that I don’t think I chose that story on purpose - so now, I’m thinking about whether and how to change that.