When Networks Need Proof and Evaluation Needs a Network

The hidden problem: we keep asking single projects to explain system change

What if the biggest mistake in social change, public policy, and philanthropy is not that we fail to measure enough, but that we measure the wrong unit?

A school program can show improved attendance. A housing initiative can count units built. A health intervention can report clinic visits. Yet the change many funders and practitioners actually want is larger, stranger, and harder to pin down: a shift in the system itself. Not just more activity, but different relationships, different incentives, different coordination, different rules of the game.

This is where two ideas often treated separately begin to collide. On one side is theory based evaluation, the discipline of asking how and why change is supposed to happen. On the other side is network based funding, the recognition that systems change rarely comes from one hero organization, but from a connected web of actors moving together.

The deeper question is not whether one model is better than the other. It is this: how do you evaluate and finance change when the thing creating change is not a project, but a living network?

That question matters because many of the hardest problems today, from climate resilience to public health to educational inequity, behave more like ecosystems than machines. They are adaptive, relational, and full of feedback loops. Treating them as isolated interventions is like trying to understand a forest by measuring one tree.

Why logic models feel reassuring, and why they often break

Evaluation has long relied on a comforting idea: if we can map inputs to outputs to outcomes, then we can determine whether a program worked. This is useful, but only up to a point. In complex environments, linear cause and effect are real, but incomplete. The same intervention can succeed in one place and fail in another, not because the intervention is bad, but because the surrounding relationships, norms, and power dynamics differ.

A theory of change helps by forcing clarity. It asks: what is the pathway from action to result? What assumptions must hold? What intermediate changes should we see first? That is already a major improvement over blind counting. But when a problem is system shaped, the most important changes are often indirect and distributed. A network might not “deliver” the solution in a straight line. Instead, it shifts who talks to whom, who trusts whom, what gets shared, and which ideas become possible.

Think of trying to improve a city’s food system. One organization can open a market, another can support urban growers, another can influence procurement rules, and another can convene local leaders. The result is not one crisp causal chain. It is a mesh of influences. If you only look for one program producing one outcome, you miss the mechanism of change altogether.

This is the first big insight: in complex systems, causality is often networked before it is measurable. The work begins in relationships long before it appears in metrics.

If the intervention is relational, then the evidence must be relational too.

The real unit of change is often the pattern, not the program

Once you accept that systems change emerges through networks, the evaluation challenge becomes clearer. You are no longer asking, “Did this program produce that result?” You are asking, “Did this network become more capable of producing results over time?”

That shift changes everything.

A network is not merely a group of organizations placed side by side. It is a pattern of coordination. Some networks are fragile and thin, held together by a few busy people and a shared label. Others are dense, adaptive, and generative. They can distribute leadership, surface information quickly, and act across multiple leverage points. In system change work, the network itself is part of the intervention.

This suggests a different theory of success. Instead of treating collaboration as a soft byproduct, treat it as a hard capacity. Ask whether the network is building:

• Shared language, so actors can name the problem in similar terms
• Trust, so information and risk can travel
• Role clarity, so efforts complement rather than duplicate one another
• Feedback loops, so learning from one node improves the whole
• Strategic diversity, so the system is attacked from multiple angles

This is where funding and evaluation meet. If you fund only discrete projects, you get discrete outputs. If you want systemic outcomes, you must also fund the connective tissue: convening, sensemaking, coordination, shared measurement, and adaptation.

The mistake many institutions make is to treat these connective functions as overhead. In reality, they are often the mechanism by which change scales.

Imagine a symphony. You could evaluate each violin, flute, and cello separately, but the music does not exist in any one instrument. The value is in the coordination. Network funding recognizes that in a complex world, the ensemble is not a support structure for the work. The ensemble is the work.

A better frame: evaluation as navigation, funding as cultivation

The usual model of evaluation behaves like an inspection: compare the final result to the plan, judge the gap, and decide whether the program succeeded. That approach assumes a relatively stable environment. But system change requires something closer to navigation. You do not simply ask whether the ship arrived exactly as planned. You ask whether it is moving in the right direction, whether it is responsive to currents, whether the crew is learning, and whether the route still makes sense given changing conditions.

This is why a theory based approach becomes especially powerful when paired with network funding. Evaluation stops being a courtroom verdict and becomes a navigation tool. It helps identify which assumptions are holding, which relationships are strengthening, and where the system is resisting change. Funding, in turn, stops being a purchase of outputs and becomes a cultivation strategy. It nourishes the conditions under which useful patterns can emerge.

That does not mean abandoning rigor. It means changing what rigor looks like.

A rigorous system change evaluation might ask:

1. What is the change theory at the level of the network, not just the project?
2. Which relationships are new, stronger, or more strategic than before?
3. What evidence shows the network is learning faster than the problem evolves?
4. Where are the leverage points, and are actors coordinating around them?
5. What unexpected effects are appearing because multiple actors are interacting?

Notice how these questions combine causality with connectivity. They do not ask only, “Did we do the thing?” They ask, “Did the system become more capable of doing the thing, repeatedly and at scale?”

A useful analogy is gardening. You can water one plant and measure its growth. But if your goal is a resilient garden, you also care about soil quality, pollination, shade, spacing, and diversity. A garden is not a pile of plants. It is an ecology. Similarly, a system change network is not a pile of grants. It is an evolving ecology of actors, assumptions, and feedback.

What changes when funders start thinking like systems stewards

Once you see the connection between theory based evaluation and network funding, a new role for funders comes into focus. Funders are not just financiers of solutions. They are architects of the conditions for learning.

This is a profound shift. Traditional philanthropy often behaves like a portfolio manager choosing promising bets. But system change funding requires something more like ecosystem stewardship. The funder must ask whether the field has enough coordination, whether the network is healthy enough to adapt, and whether evaluation is helping the actors make better decisions together.

That means funding should be designed around a few principles:

1. Fund the infrastructure of collaboration

Meeting spaces, shared data systems, backbone support, joint strategy sessions, and facilitation are not luxuries. They are the channels through which distributed action becomes coherent action.

2. Fund learning, not just delivery

The most valuable question is not only what was achieved, but what was learned about the system. Did the network discover a bottleneck? Did it identify a new ally? Did it revise a faulty assumption?

3. Fund adaptation, not just adherence

If the plan is unchanged after real-world feedback, either the problem was too simple or the team was not learning. System change requires permission to revise the theory as evidence accumulates.

4. Fund diversity of roles

Some actors advocate, others convene, others pilot, others translate. A healthy network needs different functions, not clones. Evaluation should look for complementarity, not uniformity.

5. Fund outcomes at multiple levels

Short term outputs matter, but so do intermediate relational changes and long term system shifts. If you ignore the middle, you cannot explain how change actually happened.

This is where many efforts fail. They celebrate a coalition, then evaluate it as if it were a single organization. Or they fund a program, then lament that it cannot alter a system alone. The contradiction disappears once you realize the system needs both: a coherent theory of change and a coordinated network capable of enacting it.

The practical model: three layers of evidence

If you are trying to apply these ideas, a simple framework can help. Think in three layers of evidence.

Layer 1: Direct results

These are the familiar indicators, such as enrollment, emissions reduced, or policy adopted. They tell you whether something visible changed.

Layer 2: Network capability

These indicators track whether the collaborative system is becoming stronger. Examples include:

• Number and quality of cross sector ties
• Speed of information sharing
• Degree of shared strategy
• Emergence of new leaders or connectors
• Evidence of coordinated action across nodes

Layer 3: System movement

These are the deeper shifts that suggest the problem space itself is changing:

• New norms or expectations
• Changes in incentives or resource flows
• Policy windows opening
• Power shifting toward previously excluded actors
• Unanticipated ripple effects beyond the initial intervention

The point is not to reduce everything to metrics. The point is to align measurement with the level at which change is actually happening.

A single project can often be judged on direct results alone. A system change network cannot. It must be evaluated as a living pattern of adaptation.

The more complex the problem, the more the evidence must show not only what changed, but how the capacity to change was built.

Key Takeaways

• Stop evaluating only outputs. Ask whether the network’s relationships, trust, and coordination are getting stronger.
• Treat collaboration as infrastructure. Convening, facilitation, shared learning, and backbone support are often core mechanisms of change, not overhead.
• Use theories of change as living maps. Revisit assumptions regularly, especially in complex environments where conditions shift quickly.
• Measure at three levels. Track direct results, network capability, and system movement together.
• Fund for adaptation. Build flexibility into grants so networks can respond to what they learn instead of only following the original plan.

Conclusion: the future belongs to those who can see the system and the swarm

The deepest lesson here is not that evaluation should be more sophisticated or that funding should be more collaborative. It is that the object of change has changed.

In simple settings, you can fund a project, measure its effect, and call that success. In complex settings, the real work is to alter the conditions under which many actors can learn, align, and act together. That requires a theory of how change happens and a network capable of making that theory real.

So the next time someone asks whether a grant worked, or whether a program had impact, ask a better question first: Did it strengthen the system’s ability to produce the next useful change?

That question reframes everything. It moves us from isolated wins to durable capacity, from heroic interventions to shared intelligence, from funding outcomes to cultivating the conditions that make outcomes possible. In a networked world, that may be the only definition of success that lasts.