Data Methodology & Sources

1. Primary Data Sources

All live grant opportunity data displayed on GrantMetric is sourced exclusively from official U.S. federal government APIs:

• Grants.gov API (v1/search2) — The official U.S. federal grant portal maintained by the U.S. Department of Health and Human Services (HHS). This is the primary source for all active Funding Opportunity Announcements (FOAs) across 26+ federal agencies. We query the https://apply07.grants.gov/grantsws/rest/opportunities/search/ endpoint directly.
• SAM.gov — The System for Award Management, operated by the U.S. General Services Administration (GSA), used to cross-reference agency information and award history context.
• Agency Websites — For Insights articles: NIH Reporter, NSF Award Search, SBA.gov, USDA.gov, HUD.gov, DOE Office of Science, DOD SBIR/STTR portals, EPA Grants, and agency-specific program pages are used for editorial research.

GrantMetric does not alter, anonymize, or reinterpret the raw opportunity data returned by Grants.gov. Field values (opportunity title, funding ceiling, close date, CFDA/ALN number, agency code) are displayed as returned by the API.

2. Update Frequency & Caching

GrantMetric operates on a two-tier cache architecture to balance data freshness with API rate limits and performance:

In practice, grant data displayed to users is at most 1 hour old. Grants.gov itself updates its listing approximately every 24 hours, so our 1-hour KV cache keeps GrantMetric effectively synchronized with the source. Deadlines and funding amounts shown are accurate as of the last cache refresh.

3. Data Normalization & Sector Classification

Raw Grants.gov API records contain agency codes (e.g., HHS-NIH, NSF, DOD-ARMY) and free-text category fields. GrantMetric applies a deterministic mapping to classify each opportunity into one of 5 sectors:

Classification is based on the federal agency code and CFDA category. For example: NIH and HHS opportunities → Health; DARPA, Army, Navy, Air Force → Defense; DOE Office of Science, ARPA-E → Energy; NSF, NIST → Technology; EPA, NOAA, USDA Forest Service → Environment. Opportunities that span multiple sectors are classified by primary agency focus.

4. AI Briefing Layer

The "Analyze Grant" function on individual grant pages generates a 2-sentence plain-language summary using the Anthropic Claude API (model: claude-haiku-4-5). The prompt instructs the model to:

• Summarize what the grant funds (the program purpose)
• Identify who is eligible (organizations, small businesses, researchers, governments)
• Avoid speculation about funding amounts or deadlines not present in the source data

AI briefings are generated on-demand when a user requests them — they are not pre-generated or cached. The source input for each briefing is the official grant title, description, and agency metadata returned by Grants.gov. AI-generated summaries are labeled as such and should not be used as a substitute for reading the official Funding Opportunity Announcement.

5. Insights Editorial Process

The 100+ Insights articles on GrantMetric are independently researched editorial content. Each article follows this process:

1. Source verification — Program details (funding amounts, eligibility, deadlines) are sourced from the official agency program page, the Federal Register, or the published Funding Opportunity Announcement.
2. Cross-referencing — Funding levels are cross-checked against the current fiscal year appropriations or, where announced, the agency's budget justification documents.
3. Regulatory accuracy — Where applicable, compliance requirements are cross-referenced against 2 CFR Part 200 (OMB Uniform Guidance) and agency-specific regulations.
4. Date stamping — Each article records its publication date and is updated when material program changes occur (new appropriations, revised eligibility, program closure).

Articles include specific CFDA/ALN numbers (where applicable), dollar amounts, and agency contacts so readers can independently verify all information with the source agency.

6. Known Limitations