Define the institutional need
Document the current condition, affected population, baseline performance, operational burden, and intended public value.
Define the test before implementation begins.
A controlled pilot framework for testing STEAD in a real correctional environment.
The STEAD Pilot Design and Evaluation Plan defines the problem, scope, authority, baseline, safeguards, measures, stopping rules, evaluation process, and decision gates required before a pilot can begin.
Pilot boundary: A pilot should remain limited,
reversible, independently evaluated, and
authorized by the responsible agency. It should
not create automatic approval for statewide or
systemwide expansion.
Pilot purpose
A pilot is a test of a defined claim—not a smaller version of an inevitable rollout.
A credible pilot begins with a specific problem, a measurable hypothesis, a known baseline, a defined population, and a limited operating scope.
The pilot must establish who has authority, what systems may change, what rights and safeguards remain protected, how employees will be trained, how results will be measured, and when implementation must pause or stop.
The purpose is to generate reliable evidence that can support correction, rejection, or carefully phased expansion.
01
One defined problem
Select a specific safety, fiscal, human-
development, or technology objective.
02
One measurable hypothesis
State what should improve, by how much,
for whom, and within what timeframe.
03
One bounded environment
Limit facilities, functions, users,
systems, duration, funding, and authority.
04
Independent evaluation
Separate implementation leadership from
outcome assessment wherever practical.
05
Explicit stopping rules
Define conditions that require pause,
correction, rollback, or termination.
Pilot design components
Eight components establish the complete pilot plan.
Confirm legal and executive power
Identify statutory, agency, procurement, labor, clinical, privacy, records, and oversight authority.
Bound the implementation
Limit facilities, users, functions, data, systems, vendors, duration, budget, and decision rights.
Prepare the workforce
Define staffing, training, backfill, safety, workload, labor participation, support, and change-management needs.
Protect rights and professional authority
Establish due process, privacy, access, correction, appeal, clinical control, human review, and complaint channels.
Verify technical readiness
Test identity, cybersecurity, integration, data quality, reliability, recovery, interoperability, and fallback operation.
Fund the complete pilot
Include planning, implementation, staffing, operations, support, evaluation, contingency, rollback, and transition.
Predefine the decision method
Establish metrics, data collection, comparison, review, reporting, stopping rules, and expansion thresholds.
Pilot principle
A pilot must be designed so that failure can be detected early and corrected safely.
A pilot without a baseline, stopping rule, rollback plan, or independent review is not a controlled test. It is an uncontrolled implementation with limited visibility.
STEAD pilots should preserve manual procedures, fallback operations, professional authority, continuity of care, employee safety, and the ability to restore prior operations when necessary.
The pilot succeeds only when the institution can determine what changed, why it changed, whether the result is sustainable, and whether the benefits justify the cost and risk.
Evaluation measures
Every pilot should measure performance across eight outcome categories.
Institutional security
Injuries, serious incidents, contraband, emergency response, order, and future victimization.
Employee impact
Staffing, overtime, workload, training, injuries, morale, retention, and usability.
Fiscal performance
Implementation, operations, maintenance, labor, energy, support, renewal, and total lifecycle cost.
System performance
Uptime, recovery, data quality, response, interoperability, and fallback operation.
Access and continuity
Healthcare, education, work, transportation, scheduling, completion, and continuity.
Resident outcomes
Conduct, participation, credentials, health, savings, employment, housing, and reentry.
Safeguard performance
Privacy, access, due process, correction, complaints, human review, and auditability.
Public return
Whether measurable benefits justify cost, risk, complexity, and long-term obligation.
Pilot decision gates
The pilot advances only after each decision gate is complete.
Readiness confirmed
Authority, workforce, funding, safeguards, infrastructure, and evaluation are complete.
Launch approval granted
Scope, baseline, training, acceptance, monitoring, and stopping rules are approved.
Continuation review passed
Safety, operations, cost, workforce, and safeguards remain within approved limits.
Technical and operational standards met
Reliability, usability, support, integration, security, and continuity requirements are met.
Independent analysis complete
Outcomes, costs, risks, limitations, and unintended effects are documented.
Required changes completed
Identified legal, technical, workforce, financial, clinical, or operational gaps are resolved.
Long-term operation is funded
Staffing, maintenance, support, renewal, oversight, and transition remain sustainable.
Separate expansion approval received
Broader deployment is authorized only after transparent evidence and renewed public approval.
STEAD Pilot Design and Evaluation Plan
A disciplined pilot creates evidence without forcing premature commitment.
STEAD pilots define the problem, authority, scope, workforce, safeguards, technology, funding, evaluation, measures, stopping rules, and decision gates before implementation begins.