Feasibility and co-design
$5.3m central estimateDesign legal, data, privacy, cultural, delivery, economic and evaluation architecture. Establish the full business case and pre-agreed thresholds.
Stage gates
No national rollout by interpretive dance.
The proposal moves through explicit decisions: Stage 0, a possible bounded pilot, and only then a separate question about redesign, stopping or conditional scale.
Bounded pilot
$97.6m over three yearsProceed only if Stage 0 clears its gates. Independently test implementation, participation, causal outcomes, costs, equity and harms.
Stop, redesign or scale conditionally
No automatic entitlementNational scale requires credible evidence, an affordable design, delivery capacity and public approval through the normal machinery of government.
Stage 0 exit gates
Questions that must have credible answers before a pilot.
Legal and privacy
Can consent, data separation, appeals, account governance and youth protections be designed lawfully?
Delivery
Can schools and partners operate the process without unmanageable workload, unequal access or an IT build eating the programme?
Evaluation
Can the pilot identify causal effects across universal and targeted components with adequate sample size and follow-up?
Financial
Are the pilot costs credible, the counterfactuals defensible and the evidence hurdles explicit?
Cultural legitimacy
Has the design been shaped through proper partnership rather than a Māori name being used as decorative authority?
Harm and equity
Can stigma, exclusion, coercion, gaming, industry capture and unequal school capacity be controlled?
Pilot success is not one number
A programme can lift one metric while making a mess elsewhere.
Causal outcomes
Engagement, attainment, transition, completion, employment continuity and sustained-NEET outcomes measured without double counting.
Distribution
Who benefits, who does not, and whether effects differ by need, location, ethnicity, disability, school type or pathway.
Implementation
Participation, verification burden, mentor capacity, appeals, drop-off, provider quality and actual administration cost.
Financial evidence
Observed resource use, genuine borrowing displacement, fiscal effects and value against published thresholds.
Harms
Stigma, surveillance, gaming, displacement, oversupply, wage suppression, narrowing of learning or coercive employer influence.
Stop rules
Failure is information—unless politics refuses to hear it.
The pilot should not survive because cancelling it embarrasses somebody. It should survive only if the evidence supports a defined next step.
- Redesign if the mechanism is plausible but delivery is weak.
- Narrow the target if universal delivery is poor value.
- Delay if evidence is incomplete but resolvable.
- Stop if safety, equity, causality or affordability fails.