O*NET Physical-Demands Duty Matching Explained

Why a Faxed Restriction Isn't Enough to Put Someone Back to Work

A physician's note lands on your desk: "No lifting above shoulder height; no repetitive gripping; maximum four hours standing per day." The attending provider has done their part. Now the question falls entirely to you — which of the duties you have available actually fit those restrictions, and how do you document that the fit was deliberate and defensible?

This is the daily operational problem at the center of every return-to-work (RTW) program. Getting it wrong costs time in the best case and creates a re-injury or a disputed claim in the worst. Getting it right — consistently, across multiple concurrent cases, across different supervisors and worksites — requires a repeatable method, not institutional memory or a coordinator's best guess.

ONET physical-demands data gives that method a structured spine. This article explains how the matching process works mechanically, what the ONET data actually contains and what it doesn't, and — critically — why no software tool can or should make the final duty selection automatically. By the end, you will understand how to move from a set of physician-issued restrictions to a ranked, explainable list of candidate duties, and what documentation turns that list into a defensible transitional assignment.

Includes information from O*NET 30.3, used under CC BY 4.0. O*NET® is a trademark of USDOL/ETA.

---

What O*NET Physical-Demands Data Actually Contains

ONET is the U.S. Department of Labor's occupational information network. As of version 30.3, the system covers 1,016 occupation titles and codes and contains over 19,000 occupation-specific task statements, each linked to work activity descriptors and — most relevant here — physical demands ratings. (DOL / ONET OnLine, 2025)

For each occupation, O*NET rates physical demands on a standardized scale. The relevant elements for RTW duty matching include:

• Lifting / carrying — rated by the maximum weight category the occupation requires (sedentary, light, medium, heavy, very heavy) along with frequency of lifting
• Reaching — overhead and lateral reach, rated for frequency
• Handling / fingering / feeling — manipulation demand levels
• Stooping, kneeling, crouching, crawling — postural demand frequency
• Standing and walking — time on feet during a typical workday
• Sitting — duration
• Pushing and pulling — frequency and force level
• Climbing — ladders, ramps, scaffolding

Each element carries both a level rating (how demanding the task is on that dimension) and a frequency code (never, occasionally, frequently, constantly). Those two axes — intensity and frequency — are precisely what a physician's restriction addresses. When a physician writes "no repetitive gripping," they are setting a frequency ceiling on the handling/fingering demand category. When they write "no lifting above 10 lbs," they are setting a level ceiling on the lifting demand category.

That structural parallel is what makes O*NET useful for duty matching. The occupation's demand profile and the physician's restriction profile share a common vocabulary. The matching task is essentially a filter: identify candidate duties whose demand levels and frequencies fall within the physician's stated ceilings across every restricted dimension.

What O*NET does not contain: O*NET describes occupations at a national, representative level. It does not describe the specific tasks your maintenance crew actually performs on a Tuesday, or the layout of your facility, or whether your "inventory auditor" role requires a three-rung ladder climb that isn't in the occupation description. That site-specific granularity comes from your own job demands analysis — the employer-documented record of what each position actually requires in your operation.

For an introduction to building that employer-side documentation, see our Job Demands & Essential Functions guide.

---

The Matching Sequence, Step by Step

The sequence below describes the process at a conceptual level — how a well-designed RTW duty-matching workflow moves from restriction intake to a ranked candidate list that a coordinator can act on.

Step 1 — Capture the restrictions in structured form

A restriction written as narrative prose ("avoid anything that causes pain in the shoulder") is not actionable in a systematic matching process. Restrictions need to be translated into structured fields: the affected body part, the demand category, the level ceiling, and the frequency ceiling.

This translation happens at intake, not at the moment of matching. Capturing restrictions in structured form as soon as the attending provider's note arrives prevents downstream ambiguity and creates the record you need if the assignment is ever disputed. For detailed guidance on that intake step, see restriction intake best practices.

Step 2 — Filter the candidate duty pool by demand ceilings

With structured restrictions in hand, the matching logic filters your available duties — from your task bank or duty library — against the O*NET demand ratings for the occupation codes those duties belong to.

A duty is a candidate if and only if its demand profile on every restricted dimension falls at or below the physician's stated ceiling. A single dimension that exceeds the ceiling disqualifies the duty, regardless of how well it fits on every other dimension. This is a strict logical AND, not an average or a weighted score.

Examples of how the filter applies:

Restriction	O*NET Demand Checked	Pass Condition
No lifting > 10 lbs	Lifting level ≤ sedentary/light	Occupation's maximum weight ≤ 10 lbs
No overhead reach	Reaching (overhead) frequency	Frequency code = "Never" for overhead
Max 4 hrs standing	Standing time rating	Occupation typically seated or mixed with ≤ 4 hrs standing
No repetitive gripping	Handling/fingering frequency	Frequency code ≤ "Occasionally"
No kneeling/crouching	Stooping, kneeling frequency	Frequency code = "Never"

Duties that survive the filter on every restricted dimension move forward as candidates. Duties that fail on even one dimension are removed from consideration.

Step 3 — Rank surviving candidates

Passing the filter establishes safety. Ranking within the passing set establishes fit — which candidates best match the worker's pre-injury occupation, skill set, and operational utility, while staying within restrictions.

Ranking criteria vary by program design, but common axes include:

• Proximity to the pre-injury occupation — duties that use overlapping skills have a higher training cost and re-injury risk
• Physical demand margin — how much clearance exists between the duty's demand level and the restriction ceiling (more margin is more conservative)
• Operational priority — which available duties the operation most needs covered
• Duration suitability — whether the duty can be sustained across the approved hours without a restricted dimension creeping in over time

The result is not a single automatically assigned duty. It is a ranked list of candidate duties, each with an explicit record of which restrictions were checked, which O*NET demand elements were evaluated, and how the duty's profile compared to each ceiling. That list goes to a coordinator for review.

Step 4 — Coordinator review and human selection

This step is the decisive one, and it cannot be delegated to software.

The coordinator reviews the ranked list in context: they know the worksite, the supervisor's capacity, the worker's history with the operation, whether a listed duty would require incidental tasks the job description doesn't capture, and whether the attending provider's note leaves any ambiguity that needs a call to the clinic before proceeding.

The coordinator makes a documented selection — a specific duty, a specific start date, specific approved hours — and that selection creates the evidentiary record that protects the employer if the assignment is later questioned. For a walkthrough of what that documentation needs to contain, see our light-duty job description guide.

For a broader discussion of why human review is structurally required — not just policy-preferred — in RTW AI-assisted workflows, see human-in-the-loop AI disclosure for RTW.

---

What Makes a Match Defensible, Not Just Plausible

A duty that fits the restrictions is necessary but not sufficient for a defensible assignment. Defensibility requires a documented trail that answers the question any auditor, insurer, or opposing attorney will ask: How did you determine this duty was safe for this worker on this date?

That trail has four components:

1. The restriction record. The structured intake record of what the attending provider authorized, tied to the specific note or form, with the date. See restriction intake best practices for what this record needs to contain.

2. The demand evaluation. The record of which demand dimensions were evaluated for the selected duty, what the O*NET (or employer job demands analysis) rating was on each, and how each compared to the restriction ceiling. This is the "show your work" document.

3. The human selection record. Who made the selection, on what date, based on what ranked list, and any contextual factors they weighed. This should be a named individual — not "the system."

4. The written job description provided to the attending provider. In Washington State, for example, the attending provider must approve the transitional job description in writing before Stay-at-Work (SAW) reimbursement is available. (WA L&I Complete Stay at Work Guide, 2024) A partial day counts as one reimbursable day under that program. Even in states without a formal wage-reimbursement program, sending the job description to the provider — and documenting that you did — is best practice: it gives the physician the information they need to affirm the assignment fits the clinical picture, and it creates the record that the employer sought that confirmation.

The question after a re-injury is never "did we have a transitional program?" It is "did we document that the specific duty we assigned was evaluated against this specific worker's restrictions on the day we made the offer?"

For an overview of how duty matching fits into the broader case-management workflow, see the return-to-work case management guide.

---

The Employer's Job Demands Analysis as a Necessary Layer

O*NET provides nationally representative demand ratings for 1,016 occupations. Those ratings are derived from analyst ratings and employer surveys aggregated across the U.S. economy. They are rigorous and consistent — which makes them a defensible starting point for filtering. But they describe the occupation as it is typically performed nationwide, not as your maintenance coordinator position is specifically configured at your facility.

That gap matters for three reasons:

Incidental demands. An ONET occupation profile may rate ladder climbing as "occasional" for a facilities-maintenance role. Your specific facilities-maintenance job may involve a fixed-rung ladder to an elevated parts storage area on every shift. The ONET rating is accurate for the occupation in general; your job demands analysis is accurate for the role in your building.

Equipment-specific demands. A warehousing role that ONET rates as light-lifting may involve a specific power-pallet jack model that requires grip force or arm extension your injured worker cannot tolerate. The ONET rating doesn't capture that.

Modified-duty configurations. The transitional duty you're offering may not correspond cleanly to any single O*NET occupation code — it may be a carved-out subset of tasks from two or three roles. The employer-documented job demands analysis is what defines the demand profile of that carved configuration.

Best practice is to use O*NET demand ratings as the first-pass filter and your employer-documented job demands analysis as the confirming layer. Where the two sources conflict, the more conservative (higher demand) reading governs.

To build the employer-side documentation that makes this two-layer approach work, the Job Demands & Essential Functions Workbook provides structured templates for capturing lifting, postural, and duration demands in a format that maps directly to the O*NET demand categories used in matching.

---

Common Errors That Undermine the Match

Even coordinators who understand the matching logic encounter predictable failure points. The following are the most consequential:

Matching only to the most prominent restriction. A physician note listing four restrictions gets filtered against one — the most obvious one, usually lifting. The other three are handled informally or not at all. If the assigned duty later causes a problem on a dimension the coordinator didn't formally check, the documentation gap is damaging.

Using outdated job descriptions. A job description written three years ago for a role whose physical demands have since changed — new equipment, different shift structure, added task — will produce a match that looks defensible on paper but reflects a duty that no longer exists. Job demands documentation needs to stay current.

Treating a plausible match as an approved match. The coordinator identifies a duty that seems to fit, informs the supervisor, and starts the worker — without sending the job description to the attending provider. The assignment is operationally plausible but not physician-confirmed. In states with wage-reimbursement programs, this can disqualify reimbursement; in all states, it removes a key layer of clinical accountability from the record.

Recording the outcome but not the reasoning. "Worker returned to light duty on 3/14" in the case file is an outcome notation. It is not a matching record. The documentation that matters is the reasoning: which duties were considered, why others were excluded, what restrictions were checked, and who made the final selection.

For guidance on building a task bank that supports systematic matching and avoids the outdated-description problem, see the transitional duty task bank guide.

---

From Matched Duty to Documented Offer

The matching process produces a candidate; the offer letter turns it into an assignment. What the written offer needs to contain — and the sequence for getting it in front of the right people — is outside the scope of this article, but the principle connects directly: the offer letter is where the matched duty becomes a legal communication.

For a complete walkthrough of offer-letter requirements and the documentation sequence, see the light-duty job description guide.

If your current process relies on a coordinator's judgment and a standard-form letter, but lacks the structured demand-evaluation step in between, that gap is where defensibility breaks down — not at the offer stage, but at the matching stage that precedes it.

---

Start with a System That Shows Its Work

O*NET physical-demands duty matching is not a black box. The entire value of using a structured data source is that every step in the reasoning is explicit and auditable: here are the restrictions, here are the demand ratings, here is the comparison, here is the candidate list, here is the human who made the selection on this date for this reason.

If your current matching process can't produce that chain of documentation on demand — for a carrier audit, a legal inquiry, or a state compliance review — the match may be operationally sound but it isn't program-defensible.

Transitional Duty Manager builds the demand-evaluation and case-tracking layer that makes that documentation automatic as a byproduct of the coordinator's normal workflow. The final selection is always yours. The record that you made it thoughtfully is always there.

Start a free trial and run a real case through the matching workflow.