Whether video from a correctional facility becomes usable evidence often comes down to two moments, neither of which occurs during the investigation itself.
The first is commissioning: how the system is configured, which events trigger alerts, and how it connects with other security systems. Those early decisions determine the quality and accessibility of the evidence investigators may need years later.
The second occurs when footage is exported from the recorder. That seemingly routine step is where chain of custody, authenticity, or playback issues can quietly undermine an otherwise strong case.
This article focuses on the first moment, the one most often overlooked. Commissioning may happen years before an incident and appear to be simply a purchasing or technical decision. In reality, it is an investigative decision made well before the investigation begins.
Most of what an investigation can find is decided before the system goes live
In corrections, system design is front-loaded to an unusual degree. Retention, frame rate, resolution, storage sizing, camera placement, integration with perimeter and door control, alarm behavior. All of it gets specified at project build-out and installed at commissioning. After that, the configuration tends to stay put as the project ages.
Some of that stability is deliberate. Access to change the configuration is kept narrow on purpose, so site staff generally can’t alter it, and in a corrections environment that’s a feature. That matters years later, when someone has to say exactly how the system was recording on the day of an incident, and the answer needs to be better than a guess.
The rest is friction. Configurations do get changed and alerts do get added, and a site that finds a gap can have an integrator come back and address it. That happens routinely. But it takes a service call, someone with the right access, and often a budget to authorize the work. The design can be revisited, but rarely is.
That puts real weight on who’s in the room at commissioning. Typically it’s construction, IT, and the integrator. Rarely an investigator, and almost never the person who’ll be pulling video from the system five years from now. The specification ends up making investigative decisions on their behalf.
Alarms and call-ups turn cameras into an investigation tool
Cameras record. Alarms and call-ups tell you where to look.
That is the difference between simply owning video and having a system capable of supporting an investigation. Yet alarm behavior is often one of the least developed parts of a specification. Camera counts and storage capacity receive close scrutiny because both come with clear numbers and costs. Alert logic is harder to quantify, so it is often reduced to a line item without anyone examining how it should work when an incident occurs.
When specified well, the system does the first step of the investigation for you. An intercom button gets pressed and the camera covering that spot is already on the monitor in central control. A door or perimeter event carries a timestamp that lines up against video on the same clock, so an investigator working an incident weeks after the fact can pull the event and the footage together. That’s the difference between finding the relevant ninety seconds and scrubbing hours of video hoping to catch something.
It’s worth being specific about where that event data comes from, because corrections works differently from most verticals here. It’s usually PLC-based perimeter and door control rather than a conventional access control platform. Same investigative benefit, different integration path, and a difference worth pinning down before you scope the work.
The alert set is the part worth arguing about. Facilities routinely discover, after an incident, that they wanted notification on something the system was never configured to flag, and by then the critical moment has passed. A door left unsecured is the standard example: valuable in real time, and still valuable to an investigator three weeks later, but only if someone decided at the outset that it was worth logging.
That decision has an operational tail. Call-up maps and alert logic must be maintained as the estate changes, because adding a camera isn’t finished when the camera installed and recording. If it never makes it into the call-up logic, the new camera coverage is effectively invisible to the people in the control room. It’s a strange way to lose coverage the facility just paid for.
Retention and resolution are the same decision
Retention gets treated as a policy number and resolution as a picture-quality preference. They’re the same decision, and one will silently overwrite the other.
Retention itself is rarely a site-level decision in corrections. It’s corporate governance: a 90-day baseline at a set frame rate, specified at the start of the project and locked at the administrative level so a site can’t casually change it. That lockdown exists for a reason. Without it, a well-intentioned change to the settings can break retention without anyone realizing it.
A common failure looks like this: cameras are commissioned at 4K instead of the specified 1080p, including multi-sensor units with four or five imagers, and retention drops from 90 days to under 30. Consider what that costs an investigation: any incident discovered after that window closes, and in corrections plenty are, has no video. Not video that’s hard to find. None.
Nothing about the policy is wrong in that case. A missed item on a commissioning punch list is enough to leave a facility running at a third of its required retention, often until an incident exposes the gap.
The protection against the 4K problem is headroom, specified up front. Best practice is to have integrators calculate 15 to 20 percent extra storage as a buffer. The variable is never the policy. It’s what the cameras actually produce. So check the result: actual retention against specified retention at commissioning, and again periodically. It’s a short check that protects everything downstream.
The logs you already have
Facilities routinely own investigative data nobody knows about. An investigation may need to establish whether a particular door was secured weeks earlier, and the assumption is often that there’s no way to find out. But door and perimeter control logs are rarely deleted.
In many facilities, the logging server sits in central control the entire time, capable of running date ranges. It can look like just another control computer, and the person trained on it at handover may have moved on years earlier. Nothing on these systems typically purges the log files; storage capacity is the only real limit on how far back they go. Some can hold logs for seven years.
The data is often there, sometimes in the same room where the investigation is taking place. What’s missing is the documentation needed to find and use it. That knowledge lives in one person’s memory, not in a system a shift supervisor can access when it matters.
That is the kind of information that should be built into the handover process rather than lost through employee turnover. Before recommending new technology, first determine what the site already records, where that data resides, and whether anyone still knows how to retrieve it.
Where this is heading
The interesting shift isn’t in video. It’s in what facilities do with everything around the video.
Some operators are working to move door and access control logs into cloud storage and run analysis across them, with the goal of a command center that vets alerts, decides what matters, and notifies staff when something crosses a threshold. An unsecured door, for example, would surface the moment it happens; rather than weeks later during an investigation.
That’s a meaningful change in posture, because video is inherently retrospective. A camera on a house can show how someone got in and what they took, but it won’t stop the theft.
The same holds in a facility. Staff discover an incident, review the video to see what happened, and build the evidentiary case after the fact. That process explains the event; it does not prevent it. The greater opportunity lies in the log data surrounding the video, where the signals that might help prevent the next incident often reside.
What to specify, and what to ask
If you’re scoping a system for a corrections client, four questions belong in the first conversation.
- Coverage. How much, and where are the known blind spots? Inmates learn camera coverage faster than staff do.
- Retention, asked twice. What’s the policy and who can change it? And separately, what is the system actually delivering today? Those are different questions with different answers more often than anyone expects.
- Alarms and correlation. What events call up a camera? What gets logged? Who maintains that mapping when cameras are added?
- Existing logs. What does the site already record, and does anyone know how to pull it?
All four questions focus on the facility rather than the software, which is precisely why they should be asked first. The answers reveal which capabilities matter at that specific site: how much retention is required, how many event sources must appear on a single timeline, which data points need to correlate, and what staff will need from the system on their worst day. A VMS specified around those realities will support the way the facility actually investigates.
That is the case for treating commissioning as an investigative decision rather than a purchasing one. A few additional hours at the beginning of a project can determine what a facility is able to find, reconstruct, and prove years later.
But finding the evidence is only half the job. Part 2 examines what happens after video leaves the recorder, and why the way it is exported can determine whether it holds up as evidence.
Paul Diberardino
Paul DiBerardino is the Regional Sales Manager for Salient Systems’ Southeast Region, where he partners with security integrators and end-users across multiple verticals to showcase the benefits of Salient’s video management platform. With 14 years of dedicated service at Salient and extensive experience in the security industry, Paul is known for his ability to deliver innovative, results-driven solutions that meet the unique needs of both enterprise organizations and mid-sized businesses.
Throughout his career at Salient, Paul has strengthened his expertise in sales, customer engagement, and solution delivery, positioning him as a trusted advisor to his clients. His success is built on a customer-first approach, strong technical acumen, and his commitment to cultivating long-term relationships that foster mutual growth and success.
