Definition
IMMQS measures the difference in post-move performance and retention outcomes between AI-recommended internal placements and non-AI-recommended internal moves.
The Measurement Gap
Measurement Gap and Decision Impact
It creates a composite outcome score from performance in the new role, time-to-productivity, and 12-month retention, then compares the composite for AI-matched moves against non-AI moves. A positive score means AI matching adds value. A negative score means AI-matched moves systematically underperform.
The Decision It Supports:
Should hiring managers trust the internal talent marketplace's AI matching? Internal mobility is one of the highest-stakes applications of skills intelligence: it directly affects employee careers, team capability, and organizational agility. If the AI matching engine is producing worse outcomes than manager judgment or employee self-selection, the organization needs to know before scaling its use.
Why It Requires a New Metric:
Legacy internal mobility metrics tracked volume (internal fill rate, internal mobility rate) and speed (time-to-fill for internal moves). They measured whether internal moves happened, not whether AI-recommended internal moves produced better outcomes than other internal moves. The AI-specific question (is the matching algorithm actually adding value over alternative selection methods) has no legacy answer.
The measurement gap IMMQS addresses is the absence of outcome-based evaluation for AI-driven internal talent matching: organizations are scaling AI-recommended internal moves without knowing whether the matching algorithm produces better placements than the alternatives.
Internal talent marketplaces are among the most visible AI investments in Human Capital Management. They promise to match employees to internal opportunities based on inferred skills, career trajectory analysis, and role requirements. The pitch is compelling: better matches than manager networks and job board browsing, faster internal fills, and more equitable access to opportunities. But most organizations measure the marketplace's success by volume metrics (how many internal moves happened through the platform) not by whether those moves produced better outcomes than moves that occurred through traditional channels.
This is the critical distinction. An internal talent marketplace that generates high volume but mediocre outcomes is not a success: it is automating mediocrity at scale. Conversely, a marketplace that influences fewer moves but consistently produces superior outcomes is genuinely adding value. Without IMMQS, the organization cannot tell the difference.
The outcome composite captures three dimensions that together define placement quality. Performance in the new role measures whether the employee succeeded in the position the AI recommended. Time-to-productivity measures how quickly the employee became effective, reflecting how well the AI matched capabilities to role requirements. Twelve-month retention measures whether the move was sustainable, catching cases where the AI match looked good on paper but failed in practice.
By comparing these composites between AI-matched and non-AI moves, IMMQS isolates the AI matching engine's contribution. The comparison is not trivial: selection bias is a real confounder, since employees who use the talent marketplace may differ systematically from those who find opportunities through manager relationships or self-directed search. The metric design accounts for this by controlling for pre-move performance, tenure, and role level. If the organization has reached a point where all internal moves are AI-matched (no comparison group exists), IMMQS shifts to a before-after comparison against historical non-AI move outcomes.
Without IMMQS, internal talent marketplace investments are evaluated by adoption metrics that tell the organization nothing about whether the AI is actually better at matching people to roles than the processes it replaced.
Formula
Formula:
IMMQS = Outcome_composite(AI_matched) − Outcome_composite(non_AI)
Where:
Outcome_composite = 0.4 × performance_new_role + 0.3 × (1 / time_to_productivity_weeks) + 0.3 × retained_at_12_months
Variable Definitions:
| Variable | Definition | Unit | Source |
|---|---|---|---|
| performance_new_role | Performance rating in the new role after sufficient observation period (normalized 0–1) | Normalized ratio | HRIS / Performance system |
| time_to_productivity_weeks | Weeks from role start to manager-confirmed productive contribution | Integer (weeks) | HRIS / Onboarding system |
| retained_at_12_months | Binary indicator: employee still in new role at 12-month mark | Binary (0 or 1) | HRIS |
| AI_matched | Move was recommended by the internal talent marketplace AI | Binary flag | Talent marketplace platform |
| non_AI | Move occurred without AI recommendation (manager network, self-directed, recruiter sourced) | Binary flag | HRIS / Derived |
Edge Cases and Exclusions:
- Control for pre-move performance, tenure, and role level to address selection bias
- If all internal moves are AI-matched (no comparison group), use before-after comparison against historical non-AI baseline
- Receiving team context matters: segment by team health indicators where available
- Minimum viable cohort: 30 AI-matched and 30 non-AI moves per measurement period
Data Requirements:
| Data Element | Source System | Classification |
|---|---|---|
| Internal moves with source attribution (AI-matched vs. other) | HRIS / Talent marketplace | NATIVE / DERIVED |
| Pre-move and post-move performance ratings | HRIS / Performance system | NATIVE |
| Time-to-productivity in new role | HRIS / Onboarding system | MANUAL / DERIVED |
| 12-month retention in new role | HRIS | NATIVE |
Proposed formula, pending validation pilots.