Worker

The hub object. One Worker covers employees and contingent workers, and nearly every people report resolves to it. What it holds, the data sources that expose it, and how to model it in a warehouse you own.

RuleOne data source, one primary business object, one row per instance. Pick the source to match who should count.Neverassume a headcount report counts everyone. All Active Employees excludes contingent workers by design.

◆ What Worker is— the superset of Employee and Contingent Worker, with its own identifiers.

Type	Who it covers	Identifier
Worker	Anyone with a work relationship to the company. The superset the other two roll up to.	WID plus a reference ID
Employee	People on your payroll.	Employee ID
Contingent Worker	Contractors, consultants, agency staff. In Workday, but not on payroll.	Contingent Worker ID

The split matters for every count. A person can also convert between the two types over time, contractor to employee being the common path. Model the worker once and carry the type as an attribute, so conversions read as one person with a changed type rather than two people.

◆ What hangs off Worker— job and position, compensation, organizations, events, personal data.

Cluster	What it holds	Feeds
Job & position	The position held, job profile, business title, FTE, location, manager. A worker can hold additional jobs beyond the primary.	Headcount, span of control, org roster
Compensation	Base pay, plans, grade, one-time payments, all effective-dated.	Compa-ratio, merit analysis, cost models
Organizations	Supervisory organization, company, cost center, and any custom organizations, assigned per position.	Every rollup, and the bridge to finance
Events	Hire, transfer, promotion, termination, leave. Each an effective-dated event on the worker.	Movement, turnover, retention
Personal data	Demographics, contact, identifiers. Heavily gated by domain security.	Compliance and DEI reporting, where granted

◆ The delivered windows— the worker data sources, what each returns, and the API routes.

Window	Returns	Use it for
All Workers	Every worker, employees and contingent together.	Full-population extracts. The usual base for a warehouse pull.
All Active and Terminated Workers	Everyone with a work history, current or ended. Indexed, and applies effective-dated behavior.	History rebuilds and turnover analysis. Terminated workers must be in the pull or attrition is invisible.
All Active Employees	Active employees only. No contingent workers, no terminated.	Payroll-population reports. Not a company headcount.
Employees by Organization	Employees within an organization the runner selects at a prompt.	Manager-facing operational reports, not extracts.
Get_Workers	The SOAP operation returning specified workers, or all workers when none are specified, with effective-date parameters.	Object-by-object history rebuilds. See the extraction pattern.
WQL on a worker source	Paged queries against the same data sources, over REST.	Large scheduled pulls.

Workday delivers several data sources over the same primary object so different populations and security scopes can be served. Some carry a Do Not Use designation; leave those alone. The data source, not the report, decides who can appear in the output.

◆ Reporting traps— the population, the security scope, additional jobs, and the two termination dates.

Trap	What goes wrong
The population	Two dashboards disagree on headcount because one counts contingent workers and one does not. Decide who counts once, encode it as a flag, and every report inherits the same answer.
Security scope	The same shared report returns different rows for different users, scoped by their security groups. A manager reporting missing workers is usually seeing correct security, not missing data. Extracts run as an ISU, whose grants define the full population.
Additional jobs	A worker can hold more than one position. Counting positions when you meant people inflates headcount; counting people when you meant seats understates capacity. Model workers and positions separately and pick per metric.
Sub-filters	A sub-filter on a related object decides whether the worker row appears at all; it does not narrow the related values shown for included rows. A common source of silently wrong report logic.
Two termination dates	Termination date and last day of work can differ, notice periods and pay-through arrangements sit between them. Pick one per metric and state it in the definition.

◆ How we model it— one SCD2 dimension, one event fact, one snapshot fact.

Build	Grain	What it answers
dim_worker	One row per worker per effective-dated change, with valid-from and valid-to dates and the worker type flag.	Who this person was, in job, pay, and org terms, on any date.
fct_worker_event	One row per hire, transfer, promotion, termination, or leave event.	Movement: hires, exits, internal mobility, by period and org.
fct_worker_snapshot	One row per worker per day or month. Workday does not keep this table; you build it and keep it.	Headcount, FTE, and cost trends, instantly, at any point in time.

All three carry the same conformed keys: org, job, location, cost center, and date. The cost center and company keys match the ones our Oracle and SAP models use, so people cost joins financial spend in one model. Dates follow the rules on effective dating.

Want the worker dimension built and owned?

We build dim_worker with full history, the event and snapshot facts, reconciled against Workday, and you own every line.

Talk to us

Terms on this page

Worker: Anyone with a work relationship to the company. The superset of Employee and Contingent Worker.
Contingent Worker: A contractor or agency worker. In Workday, but not on payroll.
WID: Workday ID, the internal identifier every instance carries.
primary business object: The object a data source returns one row per instance of. It dictates which related objects a report can reach.
data source: The published set of instances and fields a report or WQL query reads. Security applies here.
indexed data source: A data source pre-optimized for fast retrieval.
Get_Workers: The SOAP operation returning worker data, with effective-date parameters.
ISU: Integration System User, the read-only service account extracts run as.
additional job: A second position held by the same worker, beyond the primary.
SCD2: Slowly changing dimension, type 2: one row per change, with validity dates.
snapshot fact: A table capturing the state of every worker on each day or month.
conformed key: A dimension key shared across facts and models, so a value means the same thing everywhere.