Impact reports: how to show results donors trust

An impact report has one job: to show a funder, a board or a ministry what changed because of the work, and why they can believe it. Before you decide how to write an impact report, decide who the main reader is and what they need to do next. A programme officer preparing a renewal needs different evidence from a minister reviewing a sector.

Most readers will not read every page. Nielsen Norman Group's eye-tracking research found that people often scan in an F-shaped pattern, and it advises putting the most important points in the first two paragraphs.¹ For an impact report, that means a first page with your three or four strongest results, stated in plain sentences with numbers.

The most common weakness in donor reports is presenting activity as results. Training 400 health workers is an output. Health workers applying new skills is an outcome. Fewer avoidable deaths is impact. Readers who fund development work know the difference, and they notice when a report blurs it.

The OECD Development Assistance Committee's evaluation criteria make the same distinction. Effectiveness asks whether an intervention is achieving its objectives, while impact asks what difference it makes, meaning the significant, higher-level effects it has produced or is expected to produce.² The six criteria also include relevance, coherence, efficiency and sustainability, which are useful prompts for the questions a careful funder will ask.²

• Outputs: what you delivered, such as sessions held, materials produced or people reached
• Outcomes: what changed for people or institutions, such as skills used, services improved or policies adopted
• Impact: the longer-term difference, such as better health, income or rights

Trust comes from showing how you know, not only what you claim. For each key number, give the source, the period it covers and, where you can, a baseline or target so the reader can judge progress. If a figure is an estimate, say so. If a result came from a partner's survey, name the partner.

Bond, the UK network for international development organisations, publishes Evidence Principles and a checklist for assessing and improving the quality of evidence in evaluation reports, research reports and case studies.³ Running your draft against a checklist like this before it goes to design is a quick way to find weak claims.

Be honest about contribution. Most programmes work alongside governments, communities and other partners, so describe your part in a change rather than claiming all of it. A short section on what did not go to plan, and what you changed as a result, usually increases credibility rather than reducing it.

There is no single impact report template that every NGO must use, and many funders set their own format. Where you have freedom, the structure below works for most annual and programme-level reports. Keep the summary to one page and move detailed tables to an annex, so the main report stays short enough to read in one sitting.

• Summary page: the problem, your response and three or four headline results
• Context: the need, the people involved and the area covered
• What we did: key activities and outputs, briefly
• What changed: outcomes and impact, with evidence for each
• Voices: short, consented quotes and stories from participants and partners
• Challenges and lessons: what did not work and what you changed
• Finances and partners: how funds were used and who you worked with
• Next steps and annex: plans, methods, data tables and sources

Plain writing is part of credibility. Jargon and acronyms make results harder to check and easier to doubt. Use the words your reader uses, explain technical terms the first time they appear and write numbers as figures. Spell out each acronym once, or better, avoid it.

The US government's plain language guidance recommends expressing only one idea in each sentence, and notes that writing experts recommend paragraphs of no more than 150 words in three to eight sentences.⁴ Headings help too. Nielsen Norman Group advises starting headings with the words that carry the most information and using bullets for items in a list or process.¹ A heading such as "Clinic visits rose by a third" tells the reader more than "Health outcomes".

Good impact report design makes the evidence easier to see, not decoration around it. Choose each chart for the relationship you want to show. The Financial Times Visual Vocabulary groups charts by purpose, such as change over time, magnitude, part-to-whole and ranking, and notes that columns and bars for magnitude must start at zero.⁵ One clear chart per key result is better than a page of small ones.

Accessibility matters for a document that will be shared widely, often as a PDF and on phones. The W3C's Web Content Accessibility Guidelines set a minimum contrast ratio of 4.5:1 for normal text and 3:1 for large text.⁶ Pale text on brand colours and labels placed over photographs often fail this test.

Use photographs as evidence, not filler. A picture of a functioning water point, a classroom in use or a community meeting supports the text beside it. Caption each image with who, where and when, and only use images of people who have given informed consent.

A long PDF is rarely read in full. Plan the formats with the report, not after it: a one-page summary for ministers and board members, a web page that search engines can find, short social media cards for headline results and, where the story is strong, a short film or photo essay. Each format should point back to the full report and its sources.

In our own practice, we find that agreeing the headline results and the evidence for each before design starts saves the most time. It keeps writers, designers and programme staff working from the same facts, and avoids late changes to numbers that have already been laid out.

• Name the main reader and the decision they face
• Put three or four headline results on page one
• Label outputs, outcomes and impact clearly
• Give a source, period and baseline for each key number
• Include what did not work and what you changed
• Check consent for every photograph and quote
• Start bar charts at zero and check text contrast
• Prepare a one-page summary and a web version