Method
About this data
Source is the Department of Labor's public Form 5500 Latest datasets, not a pre-filtered FOIA extract. We keep Pennsylvania sponsor filings (state PA, with ZIP 150–196 fallback), then join Schedule A and Schedule C on ACK_ID so a missing line means it was not on the 5500.
Schedule A is carrier-contract commissions. Schedule C is plan-paid amounts to named parties. They are not added together. Amounts are annual. We do not invent EINs or compensation.
Broker identity is EIN-first: different spellings with the same EIN are one broker. Distinct EINs stay distinct, even when a name was mis-typed onto another party’s EIN in a filing. A broker is listed in search, analytics, and broker pages only when Schedule C reports a full nine-digit EIN. Name-only or truncated parties still appear on the Form 5500 filing tables when the dollars were filed, but they are not published as standalone brokers. Negative Schedule C broker fees are treated as clawbacks for analytics and broker options (not an active engagement that year) unless Schedule A also shows positive disclosed commission. Some compensation rows are therefore omitted from broker rollups. Broker options (“how’s my broker”) only include retail broker books — parties whose disclosed dollars are almost entirely Schedule C (≥90% Sch C share, or no Schedule A) are treated as captive / admin stacks and omitted from that comparison even when filed as Broker. Analytics includes broker book movement (groups and lives YoY) for listed brokers. Broker options compare placement peers by stop-loss, medical, and admin mix (similar vs dissimilar) and call out lower Schedule A or Schedule C $/life among similar peers — never summed, not a ranking.
Machine clients can use the JSON API at /api/v1 and the index at /llms.txt. Stable identifiers are sponsor EIN, broker entity ID, ACK_ID, and URL slugs.
Employer pages have three views: 5500 review, analytics (peers and flags), and broker options (similar / dissimilar placement and fee compare). Market and broker rollups also live on /analytics. The Pro /query terminal turns natural-language questions into structured group filters (stop-loss, carriers, brokers, lives, YoY premium change); results always come from filings, not model free text.