Nonprofit Financials & Filings

US nonprofits with their finances attached — revenue, expenses, assets, operating margin and multi-year growth, from IRS Form 990 filings.

ProPublica's Nonprofit Explorer holds 1.8 million tax-exempt US organisations. It exposes two things that do not talk to each other: a search that returns a directory — name, city, NTEE code, and no figures at all — and a per-organisation record that holds the money. Searching for "food banks in California" and wanting their revenue means hundreds of manual lookups.

This Actor does the lookups.

No API key. No account.

---

The four things this does that the source does not

1. It joins the directory to the money.

Every search result is enriched with its organisation record and its latest filing. That is one extra request per organisation, measured at 0.27 seconds each with no rate limiting observed — a thousand organisations takes about five minutes, and you do none of it by hand.

2. It finds the revenue that is actually there.

revenue_amount on the organisation record is populated on only 43% of organisations. totrevenue on the latest filing is populated on 95.8%. We take the best available source and record which one in revenueSource, so a figure is never silently of a different kind than it looks.

Measured on a 120-organisation run: 103 revenues came from the latest filing, 14 from the organisation record, 3 had none — 97.5% coverage against 43% from the obvious field.

3. It decodes the codes.

P210 and subsection_code: 3 are not answers. They are Human Services and 501(c)(3) Charitable, religious or educational. All 26 NTEE categories and the common 501(c) subsections are decoded, plus donationsDeductible, which is the first thing a donor needs to know.

4. It turns a stack of filings into a trajectory.

An organisation carries up to 14 filings (median 8). The source lists them; it never compares them. revenueGrowth1yPct and revenueGrowth3yPct match the same fiscal years rather than adjacent rows, because filing years skip.

Growth is left empty below a $10,000 starting base. An organisation that went from $1,000 to $1M is arithmetically up 99,694%, which is accounting noise, and it filled the top of every sort.

---

Field coverage

Measured on a 120-organisation run.

Field	Coverage
ein, name	100%
address, city, state, zipcode	99.2%
taxExemptType	96.7%
rulingDate, yearsRegistered	94.2%
totalRevenue (with fallback)	97.5%
totalAssets	85.8%
fiscalYear	85.8%
netAssets, operatingMargin	83–84%
nteeCategory	83.3%
revenueGrowth1yPct	76.7%
contributionsReceived	55.8%
officerCompensation	30.8%

---

What it is for

• Fundraising and grant seeking. Filter by sector and revenue to find foundations that actually have money to give, with their mailing address and the size of their last grant-making year.
• Selling to the nonprofit sector. totalRevenue and salariesAndWages size an organisation far better than its name does.
• Due diligence and donor research. operatingMargin, netAssets and three-year growth show whether an organisation is running a surplus or eating its reserves. 20 of 73 in one sample were running a deficit.
• Sector research. NTEE category plus revenue, aggregated across thousands of organisations.

Three dataset views ship with the Actor: Organisations, Financials and Outreach list.

---

Limits

Stated plainly, because they affect what you can conclude.

• 990 data is old, and the Actor tells you how old. Median filing age in a typical run is 3 years. The IRS publishes returns well after the fiscal year, and an organisation that merged or dissolved simply stops filing. filingAgeYears carries this on every row — never read a revenue figure without it.
• Small organisations file a postcard. Organisations under the filing threshold submit a 990-N, which carries no financial data at all. That is most of the 3% with no revenue figure; withFinancialsOnly removes them.
• Negative revenue is real. A foundation whose investment losses exceed its contributions reports negative total revenue. Those rows are not errors and are kept.
• Search results cap at 10,000. Broad queries report exactly 10,000 matches, which is the ceiling rather than the true count. Narrow by state or sector to get past it.
• One request per state. The API accepts a single state at a time, so listing ten states makes ten times the search requests.
• officerCompensationPct is dropped when negative. The source occasionally reports a negative share of compensation. A negative share has no meaning, so the field is left empty rather than filled with it.
• EIN is the join key, not the name. Nonprofit names repeat heavily across states and chapters. Match on ein.

---

Source

ProPublica Nonprofit Explorer, built on IRS Form 990 data and the IRS Exempt Organizations Business Master File. Public, no authentication. This Actor is not affiliated with ProPublica; please credit them if you republish the data.