Methodology

Automatic translation.

The rules by which we turn heterogeneous data into comparable profiles.

The scope: which organisations are included

Atlas of Aid collects non-governmental organisations for international development cooperation and humanitarian aid: those that work for the benefit of populations in other countries, especially low- and middle-income countries. It is not a list of the whole third sector.

How an organisation gets in.

• IATI: all organisations that publish data on their projects while declaring themselves NGOs (international, national, regional or partners in countries of the Global South).
• National charity registers, with a size threshold and a signal of activity abroad: England and Wales, declared activity abroad and over £1 million in income; Scotland, activity exclusively abroad and over £500,000; United States, international development and disaster relief or international human rights categories and over $10 million; Canada, over CAD 2 million and at least a quarter of spending abroad; Australia, over AUD 2 million and at least a quarter of spending on grants destined abroad; New Zealand, over NZD 1 million and international activities as the main sector, or at least half of spending abroad in at least three countries; Ireland, over €1 million and international activities, or at least three foreign countries; Sweden, where there is no public register of charities, organisations holding Svensk Insamlingskontroll's "90-account" active in Africa, Asia, South America or worldwide and over SEK 5 million in income.
• Italy: organisations registered on the list of the Italian Agency for Development Cooperation (AICS) and those present on Open Cooperazione.
• Families: members of the main NGO families (Oxfam, Save the Children, Caritas and others), reconstructed by hand.
• Verified inclusions: organisations that work directly abroad but that the automatic criteria do not capture, for example because they do not grant funds to others.
• Thematic pages: for some themes (for example sea rescue) we manually add the organisations that matter, even without a register among our sources: they enter from the European Union Transparency Register, with the budget they declare, or with the figures from their annual report, linked in the profile. We report only figures printed in the document and only for those exceeding about half a million euro in income; the others appear on the thematic page without a profile.

Whom we do not include, even when above the thresholds:

• universities, schools, hospitals and research institutes;
• religious bodies (churches, parishes, religious congregations) and organisations dealing solely with evangelisation or religious education;
• private, family or corporate foundations that limit themselves to granting funds;
• fundraising bodies whose accounts are already included in those of a group, and foreign entities registered with a US tax number;
• organisations working only for the development or services of a high-income country (according to the World Bank classification), because their activity is neither development cooperation nor humanitarian aid. Those supporting the population of a low- or middle-income country remain, even if linked to a diaspora.

Every inclusion and exclusion decided by hand is recorded with its reason. The thresholds and available registers mean that some countries (United Kingdom, United States, Canada, Australia, Italy, Sweden) are more represented than others. If you think an organisation is missing or was excluded by mistake, write to us.

1. Size comes from accounts, projects from IATI

Income, expenditure and staff come from official registers (the charities of England and Wales, Scotland, Ireland, Canada, Australia and New Zealand, the Form 990 tax returns in the United States, Svensk Insamlingskontroll data for Sweden) or from data self-declared by the organisations (Open Cooperazione for Italy, the European Union Transparency Register and, for a few organisations without a register, their annual report, linked in the profile). The profile always indicates which of the two types of data it shows. IATI, the international aid transparency standard, instead describes the projects that an organisation chooses to publish: in our checks it covers a median of about 14% of total spending. For this reason, spending on IATI projects is shown separately, labelled "declared projects", and is never presented as the size of the organisation.

2. Families are not added up without saying so

Many large NGOs are federations of national entities that transfer funds among themselves. Summing the accounts of members counts those transfers twice. When the family publishes a consolidated account or a total figure, we report it, always indicating what the figure covers and the document it comes from. When none exists, we say so.

3. Every number has a source, a year and a currency

We keep the original currency. Rankings that compare organisations from different countries convert into US dollars at the annual average exchange rate of the European Central Bank for the year of the financial period.

4. Data to be checked

Data come from registers and declarations compiled by hand: compilation or transcription errors happen. Before publication, automatic rules compare each financial year with the rest of the same organisation's series. When a figure does not add up, we flag it in the profile, in the box "Data to be checked". A flag concerns the quality of the published data, not the behaviour of the organisation.

Values we do not show (the profile indicates the year and the reason):

• out of scale: a financial year of billions in an organisation that in other years moves millions, often because of a wrong currency in the source;
• inconsistent income and expenditure: in the same year they differ by more than tenfold; we hide the one that deviates from the rest of the series (usually a typing error);
• provisional amounts: a few euro where other years have hundreds of thousands;
• incomplete return: expenditure on activities almost nil and totals below a tenth of other years, a sign that the form was only partly completed;
• mission share: if the source reports expenditure on activities almost nil (under 5%) with normal totals, the costs have almost certainly ended up in another item of the form; we remove only that share, income and expenditure remain;
• funds by source: a breakdown exceeding income by more than three times; we remove only the breakdown;
• data verified as wrong: cases checked one by one against the original documents.

Values we show with a warning:

• sharp changes: income or expenditure changing by more than tenfold from one year to the next. They are often real (a growth, an emergency, a merger), sometimes errors: we leave them visible and say so;
• IATI projects above the accounts: when spending on projects published on IATI greatly exceeds budget expenditure, we explain the most likely reason (fundraising platform, intermediary fund, publication on behalf of the family).

For founding dates that are inconsistent with the official registration, see point 5. If you know the correct figure, you can report it to us.

5. "Active since at least"

The registration date does not coincide with the founding: many organisations were born decades before registering. When the founding year is missing or unreliable, we indicate the first year documented by an official source (registration, tax recognition, entry on a public list, first publication on IATI) as "active since at least".

6. Sectors

Each source classifies activities in its own way (OECD sector codes in IATI, Charity Commission categories, US NTEE codes, sustainable development goals). We map them onto twelve common macro-sectors. The main sector is the one with the highest share; below 25% the organisation is considered multisectoral.

7. Countries

We distinguish the countries declared by the organisation in its own register from the countries with projects published on IATI. The count of IATI activities is cumulative across all years.

8. Texts and translations

Descriptions come from sources (registers, IATI) or from Wikidata. Translations into other languages are automatic and we always say so, with the original text one click away. Text pages too (such as this one, the frequently asked questions and the thematic pages) are written in Italian and automatically translated, except the Terms and the privacy notice, which have professionally curated translations; in case of doubt the Italian version prevails. When there is no text at all, the profile shows a sentence composed solely of the data.

9. Religious affiliation

For some organisations we indicate the religious affiliation, as information and not as a judgement. We distinguish three cases: emanation of a religious institution (the organisation was founded or is governed by a church or other religious institution, such as Caritas for the Catholic Church or ADRA for the Adventist Church), faith-inspired, independent (it draws on a faith but does not depend on a religious institution) and non-denominational. Where possible we indicate the denomination and the institution. Every indication is verified by hand starting from membership of religious networks, information published by the organisation and Wikidata; where nothing appears, the information has not yet been verified.

10. Beneficiaries by membership

Almost all organisations in the atlas choose whom to help on the basis of need and location. Some instead help a community defined by membership of a religious, ethnic or national group, wherever it is: for example persecuted Christians, the Coptic community, Jewish, Armenian or Assyrian communities, or Tibetans. We indicate this in the profile as information, not as a judgement, with the same criterion for every community. We do not indicate it for faith-inspired organisations that help anyone (Islamic Relief, Caritas, Khalsa Aid and many others), nor for those that support the population of a country or territory, even when linked to a diaspora. Some Armenian, Assyrian and Tibetan organisations would not fall under our sector or size criteria: we added them by hand, after reading what they declare they do, so as not to show only religious communities. The list is not complete.

11. Official checks, not scores

We do not assign reliability scores: a summary grade would hide debatable choices and would compare very different organisations. Instead, the profile reports the verifiable facts that registers and certification bodies make public: registration on the register and its status (active or removed), the last accounts filed and whether the organisation is late in filing, the reservations expressed by the auditor when the register publishes them (Australia, New Zealand), and adherence to recognised quality seals and codes of conduct (for example ZEWO in Switzerland, Svensk Insamlingskontroll in Sweden, the Core Humanitarian Standard). Every indication links to its source; the absence of a seal is not a negative signal, because many seals are national and voluntary.

12. Public funds from donors

To know how much public money an organisation receives, we use the sources of the funders, not only what the organisation declares. European Union commitments come from the Financial Transparency System; humanitarian funds from all donors come from OCHA's Financial Tracking Service; payments from 20 governmental cooperation agencies (USAID, FCDO, BMZ, Sida and others) come from the data these publish on IATI. The three sources partly overlap and must not be added together. We correct an evident error in the data: the Irish Aid amounts for 2023 and 2024 appear in thousands of euro, and we multiply them by one thousand; so the totals come close to the figures of the Irish government. The recipient is recognised by its register identifier or, if missing, by name, only when it matches a single organisation in the atlas; ambiguous names, such as that of a federation or of a UN agency coinciding with the name of a national committee, are excluded after a manual check. Not all donors indicate who receives the payments (for example Finland, Switzerland and New Zealand do not in the published data), and data for the current and previous year arrive late: the figures should be read as a minimum. Many payments go to organisations that then pass the funds on to local partners.