Intention Does Not Guarantee Impact. Goodwill Doesn’t Buy Outcomes.

From interpersonal relationships to organisations and societies, what we intend to cause and what we actually cause are rarely quite the same thing.

There is a sentence we use remarkably often when something has gone wrong: “That was not my intention.”

It is an important sentence. Intention matters when we try to understand somebody’s actions, judge responsibility or decide what should happen next. But it is also an incomplete sentence, because the absence of an intention to cause harm does not make the harm disappear. I can genuinely have meant well and still have hurt someone. My best guess is that I do it “all the time”. Explaining my intention may help you understand my behaviour; it does not retroactively change your experience of it.

There is research behind that apparently simple interpersonal observation. Adams and Inesi (2016) found across five studies that people who have caused interpersonal transgressions and those who have experienced them can be systematically misaligned in how they understand the same event. Victims were more likely than transgressors to perceive harmful behaviour as intentional, while transgressors understood their own guilt and desire for forgiveness differently. Research on apologies makes a related point: explanations and expressions of remorse can matter, but effective repair also depends on acknowledging responsibility and, increasingly, doing something about the harm caused.

That takes us quite quickly beyond the somewhat tired debate about whether intentions or outcomes matter more. Both matter, but they are different things. Intention belongs primarily to the actor. Impact emerges in the interaction between the actor, another person and the context in which the action lands.

Scale that observation up and it becomes much more interesting.

Intention meets complexity

A significant amount of management and public policy still rests, implicitly or explicitly, on relatively linear assumptions about change. We identify a problem, choose an intervention, implement it and expect an outcome. If the logic is sound enough and the implementation good enough, we assume we should broadly get what we intended.

There are many situations in which linear or direct-causality thinking is perfectly useful. If I replace a broken component in a machine with the correct new component, I can make a reasonably confident prediction about what happens next.

Human systems are different because the components notice the intervention.

They interpret it, discuss it, resist it, accommodate it, exploit it and change their own behaviour because of it. This is where complexity theory becomes useful. Complex adaptive systems are characterised by interacting and adapting agents, feedback loops, interdependencies, non-linearity and emergence. Patterns at system level can arise from local interactions without anybody having designed those patterns in advance (Lansing, 2003).

Organisations themselves display many of these characteristics. Anderson (1999) argued that complexity theory provides a useful way of understanding organisational behaviour precisely because organisations can produce surprising, nonlinear outcomes from interactions between people, structures and environments. Uhl-Bien, Marion and McKelvey (2007) subsequently brought this thinking directly into leadership, arguing that organisational capabilities such as learning, innovation and adaptation can emerge through interaction rather than simply being directed from the top.

Importantly, complexity theory is not an argument that nothing can be predicted, managed or measured. That interpretation would make it rather useless to practitioners. It is an argument for being more modest about causality.

An intervention does not travel untouched through a human system. The system responds to it, and those responses become part of what the intervention subsequently causes.

We can observe this at at least three scales: between people, within organisations and across societies. The number of actors and relationships changes dramatically as we move between those levels, but the central problem remains recognisable.

What somebody intends to put into a system is not necessarily what the system ultimately produces.

Person to person

At the smallest scale, there are two people.

A manager says something intending to reassure an employee; the employee experiences it as dismissive. Someone gives a colleague additional responsibility because they have confidence in them; the colleague experiences it as being rewarded for competence with more work. Someone provides unsolicited help because they genuinely want to support another person; the recipient experiences it as an assumption that they are incapable of doing the work themselves.

None of those examples requires malicious intent. In fact, their usefulness lies precisely in the fact that the intention can be perfectly defensible.

The impact is still real.

Adams and Inesi’s (2016) work is useful here because it demonstrates that actor and recipient can construct meaning around the same event differently. That matters particularly when we move into questions of inclusion, leadership and workplace relationships. “I did not mean it that way” can provide useful information about the intent behind a behaviour, but it cannot logically function as evidence about its impact on somebody else. The two claims answer different questions.

There is something else important here too. If our response to unintended harm is only to restate our original good intention, we learn very little. The more useful response is to become curious about the difference between what we intended and what happened.

If this is what I meant to create, but that is what you experienced, what happened between those two points?

That question creates the possibility of adaptation.

At a very small scale, that is already impact thinking.

The organisation

Now put hundreds or thousands of those people into an organisation.

The distance between intention and experience becomes considerably larger because an organisational intervention passes through multiple layers of interpretation before it becomes somebody’s reality. HR research has formalised this distinction rather neatly.

Wright and Nishii (2013) distinguish between intended HR practices, implemented HR practices and perceived HR practices. What senior management or HR designs is not necessarily what line managers implement, and what line managers implement is not necessarily what employees experience.

That should be an obvious statement, but it has fairly significant implications for how organisations talk about impact.

“We introduced a wellbeing programme” is evidence that a wellbeing programme exists. It is not yet evidence of improved wellbeing. “We have a DEI strategy” tells us something about organisational intention. “Ninety per cent of managers completed the training” tells us something about implementation.

None of those statements, by itself, tells us what subsequently changed for employees.

This is where diversity, equity and inclusion provides one of the strongest illustrations of the intention–impact problem. Leslie (2019) moves beyond the simplistic question of whether a diversity initiative “works” and instead identifies four possible unintended consequences: backfire, where progress moves in the wrong direction; negative spillover, where another undesirable effect is created; positive spillover, where benefits emerge beyond the intervention’s original objective; and false progress, where metrics improve without equivalent movement on the underlying diversity goal.

That final category should make anybody working in impact measurement slightly uncomfortable because it is by no means unique to DEI. False progress is possible anywhere that a proxy for change becomes easier to manage than the change itself.

The empirical evidence makes the organisational case even more interesting. Kaiser et al. (2013) found that the mere presence of organisational diversity structures could lead people to perceive an organisation as fairer, even where evidence indicated that members of underrepresented groups had been disadvantaged. In effect, an intervention created partly to improve fairness could also generate a signal that fairness had already been achieved.

A more recent review by Leslie, Kim and Ye (2025) reaches a suitably nuanced conclusion. Diversity initiatives can improve intended outcomes, but they can also generate backfire and negative spillover effects, and their effectiveness depends significantly on how they are designed and implemented.

The answer to that research is clearly not, “therefore, do not do DEI.” The evidence would not support that conclusion.

The more useful conclusion is that the moral legitimacy of an objective cannot substitute for evidence about the consequences of the intervention chosen to pursue it.

Goodwill does not buy outcomes.

An organisation can care deeply about inclusion and implement something that unintentionally makes aspects of inclusion worse. It can also implement something modest for one purpose and discover that it created valuable effects elsewhere.

Both possibilities become harder to see when evaluation is restricted to proving that the original plan worked.

Then we reach society

The same problem becomes larger again when we move from organisations into public policy and international development.

Here, an intervention encounters not simply employees and managers but households, communities, markets, governments, institutions, political incentives, social norms and other interventions already taking place. All of these actors and structures interact with one another, and many adapt to the intervention itself.

A programme intended to increase employment can affect wages, migration, family decisions or local labour supply. Injecting resources into a community can alter prices or power relationships. Assistance provided to one group can change how another group perceives its own position. Building institutional capacity can alter incentives inside the institution being supported.

None of those consequences needs to appear in the programme’s original theory of change for them to be real consequences of the intervention.

Robert Merton was writing about the “unanticipated consequences of purposive social action” as early as 1936, so the underlying problem is hardly new (Merton, 1936). What Dirk-Jan Koch’s work does particularly well is bring that thinking into contemporary international development and connect it more explicitly to complexity.

In Foreign Aid and Its Unintended Consequences, Koch (2023) examines multiple categories of unintended effects, including behavioural, governance, migration, conflict, price, environmental and spillover effects. Crucially, he does not present unintended effects simply as evidence that development interventions fail.

An intervention can achieve exactly the outcome it set out to achieve and simultaneously cause something else.

That distinction matters. Failure to achieve an intended objective and causing an unintended effect are not synonymous. We should be capable of asking whether an intervention achieved its intended outcomes and what else it caused.

Koch et al. (2021) tested how seriously the development sector does this by examining 644 evaluations of Dutch international development cooperation between 2000 and 2020. Only around one in six carefully considered unintended effects. Those evaluations nevertheless identified 177 unintended effects.

That finding raises an uncomfortable methodological question. If we do not find unintended effects, is that because they are absent, or because our evaluation was designed primarily to look for the outcomes we expected to see?

More recent work by Rana, Koch and Diem (2025) pushes this argument further. Drawing explicitly on complexity theory, they argue that unintended effects of public interventions should be regarded as inevitable in complex systems. Their analysis of German development evaluations is particularly useful because Germany had deliberately made attention to unintended effects more systematic. Even there, evaluation remained vulnerable to incomplete coverage, misidentification and bias.

That finding interests me far beyond international development.

If you design evaluation primarily around a programme’s original objectives, you have also designed where you will look for evidence. You become very good at answering, did the thing we expected happen? You may become considerably less good at answering the wider and arguably more important question: what changed because we intervened?

The hopeful bit: emergence works both ways

There is an understandable tendency to hear “unintended consequences” and immediately think of harm.

That is only half of the story.

Complexity creates risk because systems can respond in ways we did not predict. It also creates possibility for exactly the same reason. Beneficial outcomes can emerge that nobody designed.

This is where the parallel between the organisational and international-development literature becomes particularly striking. Leslie (2019) explicitly includes positive spillover in her theory of unintended consequences in diversity initiatives. Koch (2023) similarly draws attention to positive ripple effects in international development and argues that these are often underestimated.

An employment programme may produce confidence, social networks or changes in family expectations that ultimately prove more consequential than the employment outcome that was measured. A leadership programme may create relationships across organisational silos that become more valuable than the formal curriculum. A DEI intervention may expose another organisational issue that nobody knew to look for. A policy introduced for operational reasons may unexpectedly increase trust.

We should care about those effects for exactly the same reason that we should care about unintended harm: they happened.

And this may be where a more mature understanding of impact differs from conventional performance management. Performance management naturally wants to know whether we did what we said we would do and achieved the targets we established.

Impact thinking has to remain open for longer.

It asks what changed, for whom, how, why, what else changed alongside it and whether some of the most significant outcomes sit outside the frame we originally drew.

That is a more difficult form of accountability because it requires admitting that our theory of change is a theory, not a promise from the universe.

Integrity doesn’t scale by accident

It also requires a different kind of leadership. If people, organisations and societies are adaptive, then good intervention cannot simply be the art of designing increasingly detailed plans. It requires the capacity to observe what the system does with our plans, distinguish implementation failure from emergent effects, recognise harm without hiding behind intent, notice unexpected value without retrospectively pretending we planned it, and change course when the evidence demands it.

There is an organisational implication here that I think we talk about far too little. Values, purpose and good intentions can be relatively easy to maintain when organisations are small, relationships are direct and the people making decisions can still see the consequences of them. Growth introduces distance: more people, more layers, more interpretation, more incentives, more handovers and more opportunities for what was intended at one point in the system to become something quite different somewhere else.

Integrity doesn’t scale by accident.

It has to be designed into how an organisation, and its social programming, listens, learns, makes decisions, measures consequences and corrects itself. Not because every unintended outcome can be prevented — complexity theory tells us why that is unrealistic — but because organisations can build the capability to notice the distance between intention and impact much earlier.

That principle sits at the heart of what I am building through Amplify Impact and what I will be launching next. I am increasingly interested not just in helping organisations articulate the impact they intend to create, but in building the organisational capability required to remain accountable for what actually happens as they grow.

Because scale should not mean gradually losing sight of the people, consequences and values that made the work worth scaling in the first place.

Perhaps the most useful phrase after “that wasn’t my intention”, therefore, is not a defence at all. It is a question:

“So what did my intervention actually do?”

Good intentions still matter. I would be worried about any approach to social or organisational change that suggested otherwise. They tell us what we value and why we chose to act in the first place.

But they cannot tell us whether we succeeded.

Intention does not guarantee impact. Goodwill doesn’t buy outcomes.

In complex human systems, our responsibility does not end with what we meant to create. It extends to being curious, rigorous and humble enough to understand what we actually created — including the things we never thought to look for.

Amplify Impact B.V. works with organisations on what they actually cause, not just what they intend. That starts with measurement and reporting, and increasingly extends to the organisational capability required to stay accountable for outcomes as you grow. Integrity doesn’t scale by accident. It has to be designed in.