AI data licensing for real estate brokerages

A real estate brokerage with many affiliated agents may have few payroll employees. That distinction matters when using government statistics to assess the potential for licensing its records.

DataPayouts found 3,697 employer firms in real estate agency and brokerage that belonged to enterprises with at least 10 paid employees in 2022. They represented 2.5% of the industry's employer firms and accounted for 62.6% of annual payroll. Census SUSB 2022

A buyer would also need to examine the work completed by affiliated agents, the records retained and the rights the brokerage holds.

Company size and payroll

Census reported 150,892 employer firms and 165,585 establishments in NAICS 531210, Offices of Real Estate Agents and Brokers. Firms in enterprises with 10+ employees accounted for $21.4 billion of the industry's $34.3 billion in annual payroll.

Those firms had 17,997 brokerage establishments, approximately 4.9 locations per counted firm. Census source data and DataPayouts calculations

In 2022 U.S. real estate agency and brokerage data, the 10+ paid-employee enterprise group accounted for 2.5% of employer firms, 52.7% of paid employment and 62.6% of annual payroll. — We calculated these shares from 2022 Census SUSB data using 2017 NAICS code 531210. Each percentage uses the matching industry total. The 10+ group uses paid employment across the whole enterprise. Employee counts include part-time staff and exclude independent contractors. Businesses with no paid employees are excluded.
Sources: U.S. Census Bureau, 2022 SUSB

We calculated these shares from 2022 Census SUSB data using 2017 NAICS code 531210. Each percentage uses the matching industry total. The 10+ group uses paid employment across the whole enterprise. Employee counts include part-time staff and exclude independent contractors. Businesses with no paid employees are excluded.
Measure	Share in enterprises with 10+ paid employees (%)
Employer firms	2.4501
Paid employment	52.7429
Annual payroll	62.5714

A company with several locations may use common training, transaction-review and issue-resolution procedures. Census does not establish whether those locations share a document system or retain records of professional decisions.

The other 97.5% of employer firms may also have substantial records. Small payrolls can coexist with contractor networks and years of specialist transaction experience.

Agent numbers and paid employment

Census measures paid employees across an enterprise's establishments. Independent contractors affiliated with a brokerage are excluded from that payroll count. Census also excludes businesses with no paid employees from SUSB. Census definitions

The 2024 Nonemployer Statistics counted 819,954 businesses in real estate agency and brokerage. They reported $55.0 billion in receipts and had no paid employees. The figures count businesses, so they cannot be interpreted as a count of unique agents or brokerage-owned datasets. Census NES 2024

A brokerage assessment therefore needs information about its records as well as its staff. Relevant questions include how many transactions can be reconstructed and whether review comments remain available. Ownership of agent-created training material also requires confirmation.

A company may retain documents but lack the messages or explanations that connect them to a decision. That limits the number of complete cases a buyer could evaluate, regardless of the size of its agent roster.

In NAR's 2025 survey of REALTOR-member firms, respondents with one office reported a median of 12 transaction sides in 2024, compared with 300 among respondents with four or more offices. The difference concerns transaction activity; data quality requires a separate assessment. NAR firm profile, page 28

Employment and wage changes

BLS reported 338,220 annual-average covered jobs at private real estate agency and brokerage establishments in 2025. Employers paid $31.1 billion in wages, with average annual pay of $91,924 per covered job. BLS QCEW 2025

Employment was 9.9% lower than in 2022, while average annual pay was 4.4% higher in nominal terms. These figures allow a comparison of employment and compensation, but they cannot establish the cause of the change. BLS QCEW 2022

U.S. real estate agent and broker office payroll jobs fell 9.9%, from 375,207 in 2022 to 338,220 in 2025. — BLS QCEW reports annual-average private-sector covered jobs for NAICS 531210. These counts include part-time jobs and exclude independent contractors and unincorporated self-employed owners. A person with several jobs may be counted more than once.
Sources: BLS QCEW, 2022 · BLS QCEW, 2023 · BLS QCEW, 2024 · BLS QCEW, 2025

BLS QCEW reports annual-average private-sector covered jobs for NAICS 531210. These counts include part-time jobs and exclude independent contractors and unincorporated self-employed owners. A person with several jobs may be counted more than once.
Year	Annual average payroll jobs
2022	375207
2023	355726
2024	347691
2025	338220

Payroll includes sales, administration and transaction support. A licensing assessment needs to identify the records produced by particular tasks and whether a buyer can use them. Applying a percentage to payroll would not establish a dataset price.

Preparation costs also matter. A buyer may need consistent fields, reviewed explanations and permission to use each component. The expected payment must justify the work needed to meet those requirements.

Potential uses for transaction records

ATTOM offers commercial access to property data. In January 2026, it introduced a connection through which AI applications can retrieve structured property information. This confirms a commercial use for some property data, while leaving demand for an individual brokerage's records unproven. ATTOM announcement

A brokerage review should identify information that adds to data already available elsewhere. Possible collections include the following:

Original transaction-review examples can explain why a file was returned and how the issue was resolved.
Brokerage-authored coordination guides can document dependencies, deadlines and escalation decisions.
Training cases can pair a difficult scenario with an experienced reviewer's explanation.
Quality-review examples can show the criteria used to accept a file or request further work.

These are proposed uses that require a buyer's assessment. For example, a developer might use corrected transaction checklists to test whether a model identifies missing steps. The collection would need reliable explanations and a range of situations relevant to that test.

The developer would also need to keep evaluation cases separate from training material. Repeated blank forms would provide fewer examples of professional decisions than completed, reviewed cases.

Ownership and permitted uses

Listing photographs, MLS material, agent-created text and customer correspondence may have different owners. Permission to use them in a transaction may also differ from permission to license them for AI training. NAR's listing-content guidance addresses rights among creators, brokers and listing services.

Start by describing the systems and record categories. In a later review, the seller and buyer can identify dates, approximate case counts, export options, known gaps and who controls each type of material.

The review should check whether CRM entries, transaction documents and contemporaneous email or chat remain linked to the same decisions. A long business history may include periods with incomplete records or lost context after a software change.

A possible license can then be assessed against a specific task, such as document-review evaluation or training an assistant to complete a defined procedure. Price depends on the quality and permitted use of the records, exclusivity and continuing obligations.

Start the DataPayouts fit check with your company, systems and record categories. The form collects no client files. A detailed inventory can follow if further review is appropriate.

Data sources and limitations

Company-size findings use Census SUSB 2022 and 2017 NAICS 531210. The 2022 release remained the latest SUSB data available on October 7, 2026. We calculated the 10+ group by subtracting the under-five and five-to-nine groups from total employer firms.

The 2024 NES data use 2022 NAICS 53121, which has one six-digit child, 531210. BLS comparisons use private U.S. QCEW annual averages for 2022 and 2025.

Census employment is a March payroll headcount that includes full-time and part-time employees without conversion to full-time equivalents. Census adjusted published employment and payroll values to protect confidentiality. QCEW employment is an annual average. We report these measures separately. All dollar amounts are nominal. Archive quality and licensing value require a separate review.