The funnel, counted from events

Ask three reports how many candidates reached interview last quarter and current-state reads give three answers, because candidates moved while nobody was counting. The event fact counts everyone who ever entered the stage.

RuleCount stage entries from events over a stated window. A candidate who passed through a stage happened, whether or not they are in it now.

Neverbuild a funnel on current states. It undercounts every stage by whoever already moved on, and the miss grows with the window.

Counted from stage events, everyone who ever entered a stage counts, and the funnel stops disagreeing with the tracker.

◆ The funnel, March to May 2026entries per stage, conversion beside counts.

Stage	Entered	Conversion
Applied	620
Screened	240	38.71
Interviewed	96	40.00
Offer extended	22	22.92
Offer accepted	18	81.82
Started, ties to headcount hires	16	88.89

Each conversion is that stage's entries over the prior stage's, in percent. The bottom row is the tie: 16 starts, the same 6 plus 5 plus 5 the headcount page reports as hires and payroll prices as people. Counts travel with rates, per the same rule the turnover page enforces. Sample values are illustrative, never client data.

◆ Why the tracker disagreesstates move, events accumulate.

The delivered path reads applications' current stages through one data source, reaching transition history only through calculated-field chains that practitioners describe as gymnastics. Anyone who interviewed in March and started in May is not in interview now, so the current-state count misses them, and the miss compounds with every week of window.

The event fact has no such leak: entering a stage writes a row, and the row stays written. Pass-through rates, stage-skip patterns, and offer-decline analysis are all the same query with a different filter.

◆ The queryone grouping over the window.

The numbers above are a grouped read of the recruiting star; the schema, grain, and join keys live there.

Use case

Problem

Funnel reports disagree with the tracker and with each other because current-state reads miss everyone who already moved on.

What we build

The funnel counted from stage-entry events over a stated window, with conversion printed beside counts.

What you get

One funnel that matches the tracker, explains its own arithmetic, and bottoms out at the same starts headcount reports.

Does your funnel agree with your tracker?

We ship the event-counted funnel that ends the reconciliation meeting.

Talk to us

Terms on this page

entry: An application entering a stage. One event row, counted once.
conversion: Entries at a stage over entries at the prior stage.
window: The stated date range a funnel counts over.
current state: Where an application is now. What delivered reports read.
pass-through: The share of a stage’s entrants who advance anywhere.
calculated field: Tenant logic computed at report runtime. The delivered workaround.
tracker: Wherever talent acquisition keeps its own tally. The funnel should match it.