I have worked in humanitarian and relief programming in Yemen for the last fifteen years. During that time, I have seen major changes in how aid agencies understand and apply the principle of “Do No Harm”. Significant efforts have been made to mainstream conflict sensitivity into the way aid is delivered, through new tools, policies, mandatory analyses and donor requirements. These have strengthened capacity on the ground and improved understanding of conflict dynamics and how aid can affect them. Yet even as our capacity for conflict sensitivity grew, so too did the negative impacts that aid was having on the conflict.
In this blog, I want to reflect on that experience from the perspective of a Yemeni programme manager who has worked on the front line throughout the current conflict: what worked, what did not, and what happened when formal systems broke down. My argument is that the missing ingredient for effective conflict sensitivity and Do No Harm was never better analysis, stronger toolkits or more checklists. It was the courage to act on what we could all see was happening—together, and in time.
The Background
To understand conflict sensitivity in Yemen, you first have to understand what a decade of economic collapse did to the incentives around aid. The central bank split into two rival monetary systems, public salaries went unpaid and much of the population was left without work. Humanitarian assistance stopped being a supplement to a functioning economy and became, in many areas, one of the largest and most reliable sources of hard currency, food and basic services. Aid agencies effectively stepped into the state’s shoes, unintentionally relieving warring authorities of some of their responsibilities.
That is where the trouble began. As the formal economy collapsed, aid went from being just another resource, to being one of, if not the greatest prize worth capturing. Local factions began actively targeting our programming for co-option, looking for any route to fund their networks and legitimise their control. Frontline staff and economic recovery specialists, including colleagues I worked alongside, became direct targets — not just collateral risk from crossfire, but people deliberately harassed through intrusive travel permits, shifting checkpoint demands and intelligence monitoring designed to control our data and movements. Aid workers were routinely assaulted and branded as spies on the basis of fabricated, appearance-based accusations. Many are still in prison today, for no crime other than for having served the international aid system. Local armed groups used these tactics for two clear reasons: to pressure us into diverting resources their way, and to use humanitarian staff as political leverage.
The Hard Truth
Here is the hard truth that I learned. When war hollows out a formal economy and collapses public infrastructure, international aid stops being a neutral good and becomes a powerful variable that reshapes local power. During the many years of conflict, we as aid workers had to ask ourselves some tough questions: Were we doing more harm than good? Was our presence simply prolonging the conflict? And should donors have paused humanitarian assistance sooner than they did?
The counter-argument is obvious and important; the humanitarian imperative to save lives now should never be sacrificed lightly, and any call to pause aid risks real vulnerable people dying while agencies negotiate about principles. I don’t dismiss that risk. I lived inside it all those years. But I have also watched what happens when the humanitarian imperative is treated as absolute and unconditional over a decade: humanitarian resources become the most contested economic prize in the conflict, and armed actors organise entire strategies around capturing them.
A short, deliberate pause in aid delivery is ethically defensible when the alternative is a decade of financing the very war economy that manufactures the need we are responding to. The gravest ethical failure, in my experience, was never the temporary suspension of assistance, it was the years of quiet complicity in feeding a war economy that prolonged the suffering we claimed to be alleviating. Saving lives today cannot be the whole calculation if it also finances the killing of people we cannot reach tomorrow.
What Worked well: Local Leadership and Diplomacy
In this challenging context, I witnessed many things that didn’t work, as well as some that did. The clearest was local leadership. During repeated security lockdowns, international personnel were grounded or denied access, while national staff used their contextual knowledge and networks to keep programmes running. At a time when international managers in frontline offices often lasted only a few months, I served as a senior programme manager overseeing two offices across opposing conflict lines. It required constant diplomacy, balancing competing authorities with far less international backup than the situation warranted. It convinced me that local leadership is not a stopgap; in contexts like Yemen, it is the backbone of emergency response.
Local staff leaders also had to absorb direct political exposure. I recall one incident when a local armed group suspended our cash transfer activities and briefly detained staff. As the senior manager on site, my first concern was staff safety. In the negotiation that followed, the group demanded our full staff list and insisted their people be installed as the organisation’s primary vehicle suppliers. Rather than refusing outright and putting the detained staff at greater risk, I used calculated diplomacy. I signed a limited commitment and redirected the substantive decision upward to country-level management, buffering frontline staff from immediate retaliation and turning a dangerous field confrontation into a slower, bureaucratic procurement negotiation. It worked in that moment, but it was not a sustainable solution. Local staff can only negotiate so much on their own. Without backup from senior management, HQ and security teams, the risk in these encounters stays dangerously high.
What Didn’t Work: A Focus on Tools
Between 2015 and 2025, conflict sensitivity in Yemen evolved from a basic “Do No Harm” principle into a genuinely institutional architecture. I watched that evolution in real time. The 2016 Grand Bargain and the localisation agenda leaned into national staff and local organisations’ contextual knowledge. Whilst a good idea in principle, in practice it meant that it was people like me who were suddenly expected to carry more analytical weight, without necessarily getting more protection against the risks that this brought with it. Initiatives such as Humanitarian-Development-Peace Nexus also made sense on paper, but in the offices I managed, it mostly meant longer analysis templates layered onto teams who were already stretched thin. Donor efforts to mainstream conflict sensitivity principles by introducing increasingly sophisticated toolkits all sought to improve how aid agencies understood and responded to conflict. Each new tool was, on its own merits, a genuine improvement. None of them however, in my experience, changed what happened on the ground when a checkpoint commander decided to capture a local humanitarian staff into prison.
The result was that conflict analysis improved substantially, and localised assessments increasingly informed safer, more adaptive operations. But as the architecture improved, the violations I witnessed did not decline, and in many cases they got worse. A 2021 study found that most humanitarians believed aid was being used as a weapon of war for power and financial gain (Mosa & Fentiman, 2021). That finding matches what I saw on the ground for years: better tools in one hand, and a worsening security reality in the other.
The Real Gap
The core operational failure I witnessed, however, was not a lack of analysis or standard procedures, but it was a systemic delay in collective action. In the areas where we operated, local de facto authorities were quick to shut down any bilateral coordination that threatened their ability to divert resources. I watched procurement processes quietly captured by actors with political ties, who shaped supplier lists and tender outcomes in ways that never appeared on paper.
The system was simple and coercive: refuse to cooperate, and the project failed, permits withheld, staff detained, vehicles seized, staff accused of being unqualified. It became a constant crisis-management exercise, with interference and surveillance controlling almost every operational step. But the deeper problem was never the armed actor alone, it was that aid agencies allowed this dynamic to take hold.
Time and again, I watched as humanitarian actors, driven by the urgency of crisis response, repeatedly absorbed aid diversion in the short term at the cost of longer-term peacebuilding. Armed groups’ control over aid was rarely challenged collectively. Yet there were moments when agencies could have acted together to limit diversion and reduce conflict drivers. It wasn’t for lack of awareness or tools; it was a failure of collective will at the top. A credible threat to suspend assistance, deployed early and collectively, might have changed the trajectory before these practices became entrenched.
The aid community eventually united to pause operations in 2025, after coordinated access denials hit several agencies at once. Until then, pressure had largely been isolated and individual agencies could quietly negotiate their own way out. That difference matters, because it means the trigger for collective action wasn’t better evidence. We had comparable evidence of diversion and coercion years earlier, but the pressure had simply become too widespread for any single agency to manage alone. I believe that same collective leverage, deployed as early as 2017, would have produced real results, back when the armed actors were still consolidating power rather than firmly entrenched.
My honest reflection is this: timing is what determines whether conflict sensitivity and do no harm tools actually matter. The guidelines and toolkits only have value if decision-makers are willing to act on them at the moment when it matters most. Had the aid community made the same collective decisions in 2017 that it finally made in 2025, the outcome would likely have been very different. By 2025, the power those decisions were meant to check had already escalated far beyond what a collective pause could realistically influence.
Finally
So why did a decade of better analysis and tools coincide with worse outcomes in Yemen? Because conflict sensitivity was never primarily a technical problem. It was, and is, a political one. Analysis can tell you what is happening and why; it cannot make an aid community act collectively and on time when humanitarian principles are violated. What kept programmes alive was local leadership and calculated field diplomacy, real people making real-time judgements, often with too little institutional backup. What failed was the system’s willingness to back them with a unified and timely response.
If there is one recommendation I would offer, it is this: build the accountability mechanisms that make timely, collective action possible before the next crisis hits, not after. That was the missing lever throughout this decade, and it is the one that might have changed the outcomes I describe here.
Dr Mazen Ahmed Saif is a Yemeni humanitarian and development leader with over 15 years of program management experience in Yemen’s protracted conflict, specializing in economic recovery, and stabilization programming in hard-to-reach and frontline areas. He is the Founder and CEO of ATYAF for Studies & Management Consulting where he advises national and international NGOs on institutional strengthening, governance, conflict sensitivity, and peacebuilding and previously held senior roles with Oxfam GB, Expertise France, and other local organizations, directing multi-million-dollar programs and negotiating access with armed actors, government authorities, and international donors across Yemen’s south and north.
Note: Locations, dates, and identifying operational details in some anecdotes have been generalized to protect the safety of colleagues and partners who may still be operating in the areas described. The views expressed in this blog are those of the author.
References
Mosa, A., & Fentiman, A. (2021). Humanitarian aid & war economies: The case of Yemen. The Economics of Peace & Security, 16(1).
The Missing Ingredient in Conflict Sensitivity in Yemen: Timing, Not Tools November 13, 2026 Mazen Saif I have worked in humanitarian and relief programming in Yemen for the last fifteen years. During that time, I have seen major changes in how aid agencies understand and apply the principle of “Do No Harm”. Significant efforts have been made to mainstream conflict sensitivity into the way aid is delivered, through new…
Read more: The Missing Ingredient in Conflict Sensitivity in Yemen: Timing, Not Tools
The Metaphor of the Family Dinner: On the Importance of Social Cohesion in Moldova October 16, 2026 Dr Gal Harmat and MSocSc Gaëtan Gamba Social cohesion is often described as the glue of society: the set of relationships, of trust, power, recognition and mutual responsibility, that bind individuals, families, communities, municipalities, and identity groups together. It is what enables societies to function not merely as collections of individuals, but as interconnected wholes.…
Read more: The Metaphor of the Family Dinner: On the Importance of Social Cohesion in Moldova
Peacebuilding, stabilisation, conflict sensitivity, and social protection: time to join the dots… August 28, 2026 Rob Parker and Tim Midgley When insecurity spread across parts of Niger and Mali in recent years, humanitarian and social protection actors faced an obvious dilemma: how do you keep a cash transfer programme running when the roads aren’t safe and state presence is thin? In both countries, programme teams found workarounds. By adopting conflict…
Read more: Peacebuilding, stabilisation, conflict sensitivity, and social protection: time to join the dots…
Conflict sensitivity hubs: what works, what doesn’t and what next? August 20, 2026 Elizabeth Laruni In a world where conflicts are on the rise and becoming more complex, more humanitarian and development aid is taking place in contexts of increased fragility and insecurity. At the same time budgets are further under strain and the aid landscape has shifted, making it even more important to navigate conflict sensitivity challenges…
Read more: Conflict sensitivity hubs: what works, what doesn’t and what next?
Operationalizing Conflict Sensitivity in Iraq: From Intent to Practice July 31, 2026 Linda Atta Rashid Conflict sensitivity in Iraq is rarely absent from plans or proposals. It appears in strategies, logframes, and donor narratives, often framed as a requirement that must be demonstrated. The real challenge, however, is not whether conflict sensitivity is acknowledged, but whether it meaningfully shapes everyday decisions once programs are underway. Operationalizing conflict…
Read more: Operationalizing Conflict Sensitivity in Iraq: From Intent to Practice