The Core HR star

Two facts, five dimensions, four reports. This page is the map: every name on it links to the page that builds it.

RuleThe snapshot answers how many; the event fact answers what changed. Every report is one of those two questions.Neverbuild a table per report. Two facts and shared dimensions answer all of Core HR.

Both facts join the same five dimensions, so a number sliced by org in one report means the same thing sliced by org in another. That is the whole design.

◆ The module on one page— every table, its grain, and the pages that build and use it.

Table	Grain	Built on	Read by
fct_worker_snapshot	One row per worker per day, or per month at smaller scale.	Worker, effective dating	Headcount & FTE Trend, Span of Control, and the denominator in Turnover
fct_worker_event	One row per staffing event, classified.	Worker, effective dating	Movement, and the numerator in Turnover & Retention
dim_worker	One row per worker per change, SCD2.	Worker	All four reports
dim_org	One row per org per version of the tree.	Supervisory Organization	All four reports
dim_job	One row per job profile per change.	Job Profile	Level mix, spans by level
dim_cost_center	One row per cost center, both systems' codes.	Cost Center & Company	Headcount by budget, cost joins to the ERP
dim_date	One row per day.	Generated in the warehouse.	All four reports

◆ The proof: one series, four reports— the same sample numbers run through every report page, and they tie.

Report	April	May	The tie
Headcount	145	147	March was 142.
Movement	+3 net	+2 net	142 + 3 = 145, and 145 + 2 = 147.
Turnover	1.39%	0.68%	2 exits over 143.5 average; 1 over 146.
Span of Control	8.0	8.1	144 and 146 reporting lines over 18 managers.

This is what one model gives you: the flow explains the stock, the rates trace to the counts, and no meeting is spent arguing about whose number is right. We build client models to reconcile the same way, against Workday itself, before go-live.

Want this model live on your cloud in 10 days?

Both facts, all five dimensions, the four reports, reconciled against Workday, and you own every line.

Talk to us

Terms on this page

star schema: Facts in the middle, dimensions around them, one join away.
fact: A table of measurable rows at one grain: a day of employment, a staffing event.
dimension: The things facts are sliced by: worker, org, job, cost center, date.
grain: What one row means. Fixed per fact, never mixed.
conformed: The same dimension shared by every fact, so slices agree across reports.
SCD2: History kept as validity-dated rows, so old reports keep old truth.
stock and flow: The snapshot is the stock; events are the flow. The flow must explain the stock.
reconciliation: Proving the model's numbers against Workday's before anyone relies on them.