Internal Mobility Metrics That Hold Up: Definitions, Denominators, and the Gaming Problem
Internal mobility is the metric every organisation wants to report and few organisations define tightly enough to survive an audit. The numbers travel upward — to CHRO decks, board packs, ESG narratives — and each hop strips away a denominator. This piece lays out definitions that hold up for the three core metrics (internal fill rate, time-to-fill for internal moves, and the post-move outcome window), then catalogues the four places teams quietly game them.
Definitions that hold up
Internal fill rate. The defensible form is requisition-based: requisitions filled by an internal candidate divided by all requisitions filled, in a stated period, for a stated population (ifill = internal_fills / total_fills). Three scoping decisions must be published with the number or it is not comparable to anything, including last year’s version of itself:
- Population: all requisitions, or exempt/professional only? Hourly and contingent requisitions have structurally different internal-fill dynamics; mixing or silently excluding them changes the number.
- Backfills and restructures: is a re-org placement that never touched the market counted as an internal fill? Both answers are defensible; only one is stated.
- Candidate versus requisition view: a requisition can be “filled internally” while the internal candidate was one of forty applicants. The requisition view measures sourcing outcome; a candidate-level view (share of hires who were already employees) is nearly identical at scale but diverges for multi-headcount requisitions. Pick one and label it.
Time-to-fill for internal moves. The starting event is the entire question: requisition opened, requisition approved, or candidate identified? These produce materially different numbers for the same move, and internal moves are disproportionately exposed to the ambiguity because many begin as conversations before a requisition exists. Our baseline: clock from requisition approval to accepted offer, and separately report days_requisition_open_to_filled so pre-approval shaping is visible rather than averaged in.
The post-move outcome window. A mobility metric that stops at the fill measures process, not value. The defensible outcome set is a survival-and-performance panel on the moved cohort: retention of internal movers at 6 and 12 months against a matched external-hire cohort, and performance-rating retention — the share of movers holding or improving their rating through the first review cycle. The window matters because the signals separate in time: integration failure shows up inside six months; mis-scoped role failure shows up around the first review.
Where the metric gets gamed
None of these require bad faith. Most emerge from managers behaving rationally in a system where the metric is reported but its definition is not policed.
The wired requisition. A role is posted with the internal candidate already selected. It counts as an internal fill, fills in days, and flatters both headline metrics simultaneously while measuring nothing about mobility capability. Detection is possible in the data: requisitions with very few applicants, very short open windows, and a single interviewed candidate are a cluster worth reporting as its own category — “pre-identified internal placements” — rather than blending them into the fill rate.
Selective move classification. Upward moves are logged as mobility; lateral and downward moves are logged as transfers and vanish from the numerator. The fix is definitional: count all position changes by incumbent employees, then report the upward/lateral/downward composition as a distribution. A mobility rate with the lateral moves missing is a promotion rate wearing a mobility label.
Denominator management. Freezing external postings, or slow-walking external requisitions into the next reporting period, raises the internal share without anyone moving anywhere internal. Period-boundary effects are auditable — plot requisitions opened by week across the period boundary; a recurring trough is a reporting artefact, not a hiring pattern.
Survivorship in the outcome window. Post-move performance is reported only for movers still employed at review time, which censors exactly the failures the window exists to catch. The panel must include movers who exited, with exit coded as an outcome, not filtered as missing data.
A reporting structure that resists drift
The practical recommendation is to publish the mobility dashboard as a definitions-first artefact. Every headline number carries its scope line: population, period, denominator, clock-start convention, and censorship rules. Wired requisitions are a separately reported category rather than a quiet contaminant. The outcome panel reports the full moved cohort, exits included, and compares against a matched-hire baseline rather than a raw external-hire average — job family and level composition differ between the two groups, and an unadjusted comparison will misattribute the composition gap to the mobility programme.
What these metrics cannot tell you
Well-defined mobility metrics describe flow and early outcomes. They do not establish that internal movers perform better because they moved internally — selection is doing a great deal of work, since internal candidates arrive with observed performance histories that external candidates lack. Nor can they show whether mobility caused retention; employees who secure internal moves are, by construction, employees engaged enough to pursue them. These are correlational readings of a selection-driven process, and they should be presented as such, loudly, in the same deck.
The value of the metric stack is narrower and more useful than the usual narrative: it tells you whether the internal market is functioning — whether roles visibly open, whether insiders can actually win them, and whether the moves stick. That is a process measurement. Reported with honest denominators, it is one of the most actionable metrics in the people-analytics portfolio. Reported without them, it is a number that will eventually be quoted somewhere you cannot correct it. The data always wins over the narrative — but only if the denominator survives the journey to the slide.