Technical Deep Dive
Deployment Governance & Project Strategy
The 70/30 Project Reality
Standard PMP covers cost and quality, but in biometrics, that is at most 70% of the mission.
After 100+ biometric project deliveries, the lesson is clear: mainstream IT rules don't address the forensic stakes. When a serial offender is arrested for a minor offense, the project management of the ABIS deployment dictates whether they are linked to their past major crimes or walk free. Successful deployment requires bridging the gap between what appears to work and operational reality. Sometimes serious issues are masked and go unnoticed until somebody gets hurt.
Stewardship Lessons
1. The Knowledge Anchor
Agency staff are busy protecting the public. They rarely have the bandwidth to manage daily technical friction. Agencies need a "Knowledge Anchor," bridging the gap between agency needs and the final technical design and implementation. Industry Subject Matter Experts provide the oversight necessary to resolve these high-stakes challenges.
2. Distilling Decisions
Effective governance requires translating high-level engineering complexity into decision-ready data for busy project leads. By identifying integration friction early, the agency preserves its mission and maintains forensic-grade outcomes.
Technical Watchlist
1. The NIST Puzzle
Data isn't static. It changes with Livescan updates, JMS schema changes, and agency policy revisions. Converting this data is a "Puzzle of Black Pieces" — it requires custom use-case handling for every era of your agency's historical records that are challenging to decipher and fit together.
2. Haystack Math
Searching a warehouse (Federal DB) instead of your own haystack (Local DB) is a trade-off that abandons the 60-70% recidivism rate that has a real impact on solving crimes with a local ABIS. Governance ensures the local system is tuned to catch the difficult prints that Fed systems might miss.
A Technical Escalation Saves Lives
Why forensic curiosity outweighs excuses.
A repeating issue in biometric deployment is the assumption that if a simple set of matcher tests work, the system is fully functional. In one landmark case, a customer escalated a recurring accuracy issue that had been dismissed by a technician because the system "passed" a basic check.
By refusing to accept a superficial "it’s working" status and doing a deep dive into the system, we identified an issue hidden beneath the surface. Once the technical resolution was deployed, the agency lead shared that they immediately identified a serial rapist who had been missed earlier. This illustrates the real-world impact of deployment governance: you need a team that understands the industry, not just the interface, to ensure the mission never fails.
Many years ago, I had a customer (when I worked for a technology vendor) escalate an issue to me that was initially dismissed. It was not easy to see, but the customer knew something was wrong. I worked with the technical team to resolve the issue. Shortly after the resolution, the agency lead called me to share that they had just identified a serial rapist they missed before. This is the real-world impact of how complicated these systems can be and the deep subject matter expertise needed to manage these programs.
"The most dangerous failure is the one that looks like a success.
Don’t let a Technical Illusion
Compromise your forensic mission."
Engineering the "Go-Live"
Tenprint Validation
During repository builds, search every record against the full DB. This validates tenprint accuracy meets or exceeds the previous system and often finds matches missed by human prior human or system error.
Latent Sampling
Manual review is non-negotiable. Sample historical idents and no hits. Make sure system performance is similar to your prior system. *Note* poor quality latent accuracy differences may not be an issue.
Conversion Integrity
Migration is just moving IDs. Conversion is transforming data. We advocate for exhaustive spot-checks to ensure minutiae mapped correctly, all biometrics are present, data fields are mapped, etc.
The Lifecycle of Acceptance
Ensuring forensic stability from the lab to the street.
FAT / SAT
Technical Baseline: Proving the code is stable and hardware is connected. Standard IT delivery.
UAT
User Acceptance: Confirming staff can navigate workflows and the system handles known agency search scenarios.
Go-Live Validation
Production Cut-Over: Final audit of conversion data and matcher settings in the live environment to ensure no forensic drift occurred during migration.
OAT: Operational
The 90-Day Burn-In: Measuring performance against mission objectives after months of real-world volume. Final safeguard against the Technical Illusion.
Strategic Frontloading & The RTM
A mission-critical deployment involves thousands of technical and forensic requirements. We utilize a Requirements Traceability Matrix (RTM) to manage these objectives from FAT through OAT.
Our governance strategy is to "Push Left"—frontloading validation as early as possible. While final matcher accuracy requires the high-volume operational load of OAT to surface complex corner cases, frontloading core technical tests ensures that the "Technical Illusion" doesn't survive into production.
Stewardship Insight: Subject Matter Expertise led governance ensures the vendor stays through the OAT phase to prove that "Live" actually means "Solving."
Bridging the Custom Gap
Commercial Off-the-Shelf (COTS) solutions often leave meaningful gaps in local and state policy compliance and limit key intel that impacts officer and public safety. Some examples of important customizations are:
Juvenile Records
Handling sensitive records requires more than a field; it requires workflow logic that matches local and state mandates and eliminates duplicate or corrective actions resulting from unoptimized processes.
MobileID Integration
MobileID can be a simple Hit or No Hit response. The power of MobileID is with the smart integration. Do you include a Mugshot, Want/Warrant, Know Dangerous Persons Identifier, etc.
Jail Management
When done correctly, interfaces between systems like JMS/RMS and Livescan can eliminate duplication efforts, strengthen data integrity, and improve officer safety. This lets the officer focus on the subject, not the technology.
The Silent Failure: Missing Search Workflows
I have visited multiple agencies where mission-critical workflows were completely missing from their "new" systems. This often happens because vendors have high turnover and training focuses on basic usage rather than forensic needs, or there was a misalignment on scope during the sales process.
For example, without enrolling unsolved latents and searching them against every new booking—and searching unsolved against unsolved—the system is only finding "obvious", current point in time, idents. ABIS systems need to be configured, and examiners need to be trained, to account for all workflows. This can result in catchable serial criminals remaining at large.
If the search logic isn't forensically curious, the identification is lost before it begins.
Walt Stelz
BCP CEO
"The 30% gap is where
identifications are won or lost."
GOVERNANCE IS THE BRIDGE BETWEEN VISION AND REALITY.