The data behind the predictions
Not Just More Data.
Better Data.
Anyone can ingest IRS filings. We resolve, clean, filter, and validate every grant so the data actually tells you something useful.
The SciRise Foundation Intelligence Dataset
Every number represents verified, entity-resolved, and connected data.
4.8M+
Grants Analyzed
Every grant entity-resolved to a verified nonprofit
16,000+
Foundations
Actively distributing $1M+ annually
300,000+
Nonprofits
Matched and verified across filings
$200B+
Annual Giving
Total foundation giving tracked
670K+
Funding Relationships
Verified foundation-to-nonprofit connections
370K+
Foundation Contacts
Officers, directors, and trustees
Nationwide
Geographic Coverage
All 50 states and US territories
4 Years
Filing Depth
Multi-year patterns, not single snapshots
All data sourced from IRS 990 and 990-PF filings. Updated as new filings become available.
The Problem With Raw 990 Data
Most foundation data tools stop at ingestion. They import IRS filings and present them as-is. That creates real problems.
Unresolved recipient names
"Duke University" appears as 50+ different text strings across IRS filings. Without entity resolution, each variant looks like a separate organization.
Intermediaries counted as funders
Donor-advised funds and fiscal sponsors appear as grant-makers, inflating foundation counts and obscuring who actually makes funding decisions.
Single-year snapshots
One year of data misses patterns. A foundation that consistently funds health research looks identical to one that made a single one-off grant.
No validation
Without testing against real outcomes, there's no way to know if the data actually helps you find funders or just looks impressive.
What We Do Differently
Every grant in SciRise goes through a multi-stage quality process before it reaches you.
Ingest IRS Filings
We start with multi-year IRS 990 and 990-PF filings, the authoritative public record of foundation giving in the United States.
Resolve Every Grant
Raw filings list recipients as free-text names. We match every grant to a verified nonprofit using multi-stage entity resolution, not just string matching.
Remove Intermediaries
Donor-advised funds, fiscal sponsors, and regranting organizations inflate grant counts without representing real funding decisions. We identify and remove them.
Validate Against Reality
We test our data against real funding outcomes using holdout years the model never saw. Every claim we make is backed by measurable results.
Processed Data vs. Raw Filings
The difference between ingesting data and understanding it.
| Capability | SciRise | Raw 990 Ingestion |
|---|---|---|
| Entity resolution | ||
| Intermediaries filtered | ||
| Grant-to-nonprofit matching | ||
| Validated against real outcomes | ||
| Every recommendation traceable to source filing | ||
| Predictive model (not just search) |
Statistically Validated
We trained on historical filings, then tested whether we could predict new funding relationships that actually appeared in a future year. The model never saw the test data.
28.7%
Hit rate in top 10
Nearly 1 in 3 nonprofits found a real future funder in our top picks.
13.6x
More accurate than keyword search
Head-to-head comparison on the same data and evaluation criteria.
30,000+
Nonprofits tested
Statistically significant (p < 0.0001).
Transparent and Traceable
Every recommendation links back to real IRS filings. No black boxes.
Source-linked recommendations
Every foundation recommendation traces back to specific IRS filings and real grant history. You can see exactly why a foundation was surfaced.
Multi-year patterns, not single snapshots
We analyze multiple years of filings to distinguish sustained giving from one-off grants. This means more reliable signals and fewer false leads.
Explainable methodology
Our validation methodology, test design, and results are published openly. We believe data tools should be held to the same standard as the research they support.
See What Better Data Looks Like
Sign up and explore foundation recommendations built on verified, validated data.
No credit card required.