
Unpack the file and keep a copy
Every pack downloads as a gzip-compressed CSV. Extract it with your system's archive tool, or gunzip on the command line, and keep an untouched copy before you change anything. The agent file has 25 columns and the adjuster file 27; an empty cell means the record has no value for that field. The CSV documentation describes every column.
Keep identifiers as text
Spreadsheet apps guess at data types and can strip leading zeros, so import npn, license_number, postal_code and business_phone as text. Postal codes in New England start with a zero, and an NPN is an identifier, not a number. Size matters too: an Excel worksheet holds at most 1,048,576 rows, fewer than the 1,755,281 in the nationwide agent file, so open that file in a database or a data tool, or use the ZIP of state files that comes with a nationwide order.
Map the columns
Map business_email to the contact's email, business_phone to phone, and first_name and last_name to the name fields. agency_name belongs in the company field, and address, city, state and postal_code fill the address. Keep npn, license_type, lines_of_authority, license_status, expiration_date and years_licensed as custom properties, and keep ingested_at so everyone can see how old the record is. The is_life_health and is_property_casualty flags work well as yes-or-no properties for building lists.

Size each import to your CRM
HubSpot's paid plans take up to 1,048,576 rows and 512 MB per file, so any state pack goes in as one file and the nationwide agent file, 1,755,281 rows, as two. HubSpot's free tools stop at 20 MB per file and 500,000 rows a day, which at this file's width is roughly 60,000 rows per file.
Salesforce's Data Import Wizard takes up to 50,000 records at a time. 46 of the 51 agent packs fit in a single import; Florida, with 397,516 records, would take 8, and for files that large Salesforce recommends its Data Loader instead.
Pipedrive takes up to 200,000 rows per import, in a file under 50 MB. A row in these files averages about 300 bytes, a little more in the adjuster file, so the 50 MB limit usually comes first: split large packs into files of 120,000 rows, which puts Florida into 4 imports. Pipedrive also needs a name for every person, so map first_name and last_name.
These are the limits each vendor published in October 2026. Check the current ones before a large import.

Decide how duplicates are matched
HubSpot matches contacts on email, and a contact's email must stay unique even if you add your own unique ID such as npn. Some agencies file one inbox for several agents, so decide whether those agents share one contact or go in without an email. Pipedrive offers to merge the duplicates it detects. Whatever the CRM, decide on the matching key before the first import, not after.
Importing the adjuster file
The adjuster file maps the same way, with business_name in place of agency_name. Multi-value columns such as license_states and lines_of_authority are joined with a semicolon and a space; keep them as text or split them into multi-select values. Check email_shared_count before you match on email: 1 means the address belongs to that adjuster alone, and a larger number means a shared licensing inbox.
Test a small batch
Choose a few representative rows, including a row with a missing email and one with several lines of authority. Map the fields with your CRM's current import tool, then review the duplicates and the formatting before you import the rest.
Use the data with care
A publicly reported contact address is not evidence of marketing consent. Apply your own suppression rules and recipient preferences. Separate research and recruiting workflows from any automated campaign, and keep the snapshot date attached to the records.


