What This Covers
This hub is for staff at nonprofits and other organizations who have to plan a new program. It runs from understanding the problem through to objectives and an action plan, and stops before implementation. Design means more than writing a document. You find out what problem to solve, work out who else is involved, look at what interventions have been shown to work, and take an honest look at your own resources and skills. The tasks below follow the usual order of that work.
Start Here
Work roughly in this order. Where conflict is a risk, conflict sensitivity should inform all levels of an intervention and every stage of the programming cycle, so run the harm check (the last item below) alongside every step, not only at the end.
- Assess needs. Ask community members what they see as the most important needs, using a survey with enough reliable information to act on. Run a needs assessment walks through it.
- Map stakeholders. List government, beneficiaries, other NGOs and key individuals, and what each contributes.
- Build a problem tree. Put causes and effects on paper with your team and community representatives, then check them against data. Reversing its negative statements into positive ones is one way to find possible goals and objectives. See problem analysis, and use the problem tree prompt for a first draft from your notes.
- Research what works. Find out which interventions have been shown to work. This is one of the most critical steps and the one most often missed. If there is no prior evidence for a new approach, pilot it before scaling up.
- Check your own resources and skills. Look honestly at what your organization can deliver. There is no point designing an intervention your team has no specialists for.
- Build a theory of change. Map the sequence of change and the assumptions behind it. Theory of change for grant applications shows how to present one to a funder.
- Write a logframe. Connect activities, outputs, outcomes and goal, and test that the levels hold together.
- Set objectives and an action plan. Turn goals into measurable objectives, then list each step, who does it and by when. The project design document template brings it together.
- Plan the exit. See exit strategy for how to think about ending or handing over the work.
- Check for harm. Where conflict is a risk, run a do no harm analysis, with gender included, and update it during implementation.
Key Concepts
- Theory of change. A map of how change happens, from long-term goal back through preconditions, with assumptions, interventions, indicators and a narrative.
- Logframe. A table of activities, short-term outputs, medium-term outcomes and a long-term goal. There is no agreed standard, so the top level may be called goal, outcome, objective or impact.
- Assumptions. What must be true for one level to lead to the next. State them so they can be tested.
- Indicators and means of verification. Indicators show whether a level was achieved, and the means of verification says how each will be measured.
- Objectives. Specific, measurable results saying how much of what will be done by when, ideally SMART: specific, measurable, achievable, relevant and time-bound.
- Action plan. The list of steps toward the objectives: what happens, who does it, and by when.
- Do no harm. An analysis of how assistance might worsen conflict, built around dividing and connecting issues and actors.
Templates
The project design document template is where the design comes together. Build these four as you go.
- Stakeholder table. Use when mapping who is involved. Columns: stakeholder, key problems, motivations, potential contribution.
- Problem tree worksheet. Use early in the design. Core problem in the middle, causes as roots, effects as branches.
- Logframe. Use for proposals and evaluation. Levels, indicators, means of verification, and risks and assumptions.
- Action plan. Use after objectives are set. For each step: what will happen, who, by when and for how long, resources, and who should know.
AI for This Topic
In program design, an AI assistant can help with drafting content, summarizing information, generating ideas and supporting research and analysis, for example when drafting a logframe or summarizing evidence on an intervention. The risk lies in the data you put into it, so the rules you already have against uploading certain information to outside services apply here too. A simple test, from nonprofit AI policy guidance: if you would not email it, do not paste it into an AI tool. Apply that test to material such as raw needs assessment responses or confidential internal strategy before you paste it in. Staff should also verify anything the assistant produces for accuracy and bias, and review it before it is shared or used to make a decision.