Independent contractor classification audit: start now, not later
The Labor Department's proposed rewrite of the employee versus contractor test is still pending, and a reopened self-reporting lane gives HR teams a narrow, cheaper window to fix classification records first.
An independent contractor classification audit is one of those projects HR teams keep on the someday list, usually because the legal answer was settled once, years ago, at contract signing. That window is about to get more expensive. The Labor Department's Wage and Hour Division proposed a rule in February 2026 clarifying employee versus independent contractor status under the Fair Labor Standards Act, and as of now it has not been finalized. The interim period is the opportunity.
What the DOL proposal actually changes for people ops
The February 26, 2026 news release from the Department of Labor framed the proposal as a clarification of employee versus independent contractor status under the FLSA. Employer-side firms treated it as significant almost immediately: PilieroMazza published analysis the following day, Jackson Lewis in early March, and Butler Snow later that month. When three defense-side practices issue client guidance within a month of a proposal, that is a signal about expected enforcement salience, not just academic interest.
The more consequential detail came from Dechert in June 2026, which described the department as advancing independent contractor and joint employment rulemaking in tandem. That pairing matters operationally. Classification determines whether a worker is covered by FLSA wage and hour obligations at all. Joint employment determines who is on the hook when the worker is engaged through a staffing firm, a vendor, or a platform. Moving both at once means an employer can see its exposure shift on two axes in the same quarter, even if nothing about the actual working relationship changed.
Coverage from Sommers Schwartz and a subsequent where-it-stands explainer indicate the rule remains unfinalized. Pending is not the same as dormant. It means employers are currently being judged against one standard while preparing for another, and any documentation gap they have today will be evaluated later under whichever test survives.
Classification is not a legal question answered once at signing. It is a recordkeeping obligation people ops owns every quarter.
The self-reporting lane is open while the standard is being rewritten
HR Dive reported in January 2026 that federal officials were promoting a reopened and expanded wage-and-hour violation self-reporting program for employers. Read alongside the pending rulemaking, that is the whole story: a voluntary correction path is available during a rewrite of the very standard employers will eventually be measured by.
Self-reporting programs reward a specific capability, and it is not good intentions. It is the ability to produce clean, contemporaneous evidence of how a classification decision was made, who made it, what facts were considered, and when the relationship was last reviewed. Employers who can assemble that file quickly get to control the narrative and the timeline. Employers who cannot end up reconstructing years of reasoning under adversarial conditions, usually in litigation, usually with a plaintiff's counsel selecting which emails matter.
SHRM also flagged in January 2026 that FLSA opinion letters issued in 2026 function as a distinct interpretive signal. Opinion letters are narrow by design, but they tell compliance teams which fact patterns the agency is thinking about. Tracking them costs almost nothing and gives people ops a defensible reason to reprioritize which contractor populations get reviewed first.
Why federal and state tests will not move together
ADP SPARK counted dozens of state-specific HR compliance changes taking effect in 2026. State classification tests have never been synchronized with the federal standard, and a federal rewrite does nothing to align them. Several states apply stricter multi-factor or ABC-style analyses that will not budge because the FLSA test shifts.
The practical result is uncomfortable for anyone who manages a distributed contractor population. The same engagement terms, the same statement of work, the same weekly hours can be defensible in one state and misclassified in another. If your contractor roster is managed centrally but your workers sit in fifteen jurisdictions, you do not have one classification question. You have fifteen, and the federal proposal only touches the floor beneath all of them.
This is where the ownership problem shows up. Classification is usually treated as a legal question answered once. It is better understood as a recordkeeping obligation that people operations owns continuously, because the facts that drive the answer are HR facts: scheduling control, exclusivity, duration, integration into core operations, who supplies tools, who sets rates.
Building the classification evidence file before the rule lands
Most HR teams cannot produce a classification evidence file on demand because the reasoning does not live in the HRIS. It lives in procurement systems, in legal's contract repository, and in email threads between a hiring manager and a business partner from three years ago. None of that is searchable in a way that survives an audit request with a two-week response window.
A workable starting point is narrow. Inventory every non-employee worker paid through accounts payable, not just those flagged as contractors in the HRIS. Map each one to a jurisdiction and a governing test. Capture, in one structured field, the factual basis for the classification and the date it was last reviewed. Flag any engagement running longer than twelve months, any worker with a single client relationship, and any role that also exists on the employee side of the house. Those three flags surface most of the real risk.
Then decide who owns re-review. Annual is defensible. Event-driven is better: renewal, scope change, rate change, or a shift in where the worker sits. The point is not to reach a perfect answer under a standard that does not exist yet. The point is to be able to show, later, that the question was asked deliberately and answered on the facts available at the time.
