A 12-location retail chain: cameras that finally get watched before something happens, not after
Each store ran its own disconnected DVR system with no shared visibility. Merilsoft's View Scan product line put every site's cameras and door access into one dashboard.
- Incident footage review
- Store-by-store phone calls replaced by one searchable dashboard
- Alert noise
- AI person/vehicle detection over raw motion cut false alerts sharply
- Stockroom access
- Shared physical keys replaced by logged, role-based entry
Each of this retail chain's 12 stores ran its own DVR, bought and installed whenever that location happened to need one, with different brands, different retention windows, and none of them talking to head office. Footage only got reviewed after something had already gone wrong, and stockroom access still ran on physical keys re-cut for years of departed staff.
Merilsoft rolled out its View Scan product line — Gateway Manager, Protect, and Entrance Manager — store by store over several weeks, starting with the two locations with the most incident history. Store-by-store phone calls to pull footage were replaced by one searchable dashboard, AI person and vehicle detection cut false alerts sharply, and shared physical keys gave way to logged, role-based badge entry.
Client overview
The chain operates 12 retail locations under a centralized loss-prevention function, with each store historically responsible for its own security equipment purchases — a pattern common among retailers that grew store by store rather than under a single infrastructure standard from day one.
By the time this engagement started, the loss-prevention team's actual working process was a phone tree: a call to whichever store had the incident, a wait for someone on-site to locate the right DVR and pull the right timestamp, and an export that was often too low-resolution to actually be useful.
The challenge
Twelve DVRs, twelve different systems, no shared view
Each of the 12 stores had its own DVR, bought and installed whenever that location happened to need one — different brands, different retention windows, none of them talking to head office. When an incident happened, loss prevention had to call that specific store, wait for someone to pull the right date and time, and export a clip that half the time was too low-resolution to be useful.
Footage nobody watched until after the fact
Because nobody was watching feeds live, footage only got reviewed after something had already gone wrong — a shortage at count, a complaint, a claim. By then the same blind spot had usually been exploited more than once.
Stockroom keys nobody could account for
Stockroom and back-office access ran on physical keys, several of them cut and re-cut over the years for staff who'd since left. There was no record of who'd been in a stockroom at 2am, only whoever happened to still have a key.
Project objectives
- Put every store's live and recorded camera feed on one dashboard head office can pull up without a phone call.
- Cut false alerts by detecting actual people and vehicles instead of raw motion.
- Replace shared physical keys with role-based badge access, logged by name and timestamp.
- Validate the workflow at the highest-incident stores before rolling out chain-wide.
Discovery & business analysis
Merilsoft's discovery started with the loss-prevention team's actual working process rather than a description of it — sitting with the team through real footage pulls and watching the phone tree play out end to end. A call went to the store, someone on-site went looking for the right DVR and the right timestamp, and the clip that eventually came back was often too low-resolution to settle anything. What those sessions established was that the problem was never the cameras themselves — it was that head office had no way to reach any of them without asking a store to do it first.
From there the work moved to the stores themselves. Merilsoft walked each of the 12 locations and catalogued what was actually installed: the DVR brand, its retention window, and whether the cameras attached to it had any network path at all. Some were recent enough to integrate directly; others were older analog units with nothing to integrate to — a spread wide enough that no single approach was going to cover all 12 sites. The same walk mapped every stockroom and back-office door and confirmed what the team already suspected: nobody could say how many keys were in circulation or who still held them.
The sequencing decisions came out of discovery too. Merilsoft went through the loss-prevention team's own incident history and found it wasn't spread evenly across the chain — it concentrated at a couple of locations. Those two became the pilot sites, on the reasoning that alerting rules tested against a real workload would surface problems a quiet store never would. Watching the stores open and close surfaced the other timing decision: badge issuance for existing staff would have collided with opening and closing routines if it landed alongside the camera work, so Entrance Manager was scheduled to follow a few weeks behind cameras at each site instead of going in with them.
Solution architecture
One dashboard, every site's cameras
Merilsoft rolled out View Scan Gateway Manager to bring every site's cameras onto one dashboard, so head office can pull up any store's feed — live or recorded — without a phone call. View Scan Protect adds AI person and vehicle detection, so an alert means an actual event instead of a raccoon setting off motion sensors at 3am.
Badge access that replaces the physical key
View Scan Entrance Manager replaced the physical key system with role-based access — each employee's badge opens what their role needs and nothing else, and every entry is logged with a name and timestamp. Revoking access for someone who's left is now a two-minute change instead of re-keying a lock.
Rollout led by the highest-incident stores
The rollout ran store by store over several weeks, starting with the two locations that had the most incident history, so the loss-prevention team could validate the workflow before it went chain-wide.
Technology stack
| Component | Technology | Why |
|---|---|---|
| Video management | View Scan Gateway Manager (cloud VMS) | Every site's live and recorded feed accessible from one dashboard |
| AI detection | View Scan Protect (edge AI) | Person and vehicle detection means alerts fire on real events, not raw motion |
| Access control | View Scan Entrance Manager | Role-based badge access with every entry logged by name and timestamp |
| Storage | Distributed edge NVR + cloud backup | Footage exportable on demand without a phone call to the store |
| Alerting | Push notification service | Loss prevention gets a real-time alert instead of finding out after the fact |
Key features
Unified camera dashboard
Before
Pulling footage meant calling a specific store and waiting for an export
Now
Any site's live or recorded feed is available from one dashboard
Store-by-store phone calls replaced by a searchable system
AI person & vehicle detection
Before
Raw motion alerts were noisy and mostly ignored
Now
Alerts fire only on detected people and vehicles
False alerts cut sharply, real events actually get attention
Role-based badge access
Before
Shared physical keys had no record of who entered when
Now
Each badge opens only what its role needs, logged by name and time
Revoking a departed employee's access is a two-minute change
Phased, incident-led rollout
Before
A chain-wide cutover risked disrupting all 12 stores at once
Now
Rollout started at the two highest-incident stores to validate the workflow
Process proven before it reached the rest of the chain
Rollout & delivery
Rollout began at the two stores with the heaviest incident history, giving the loss-prevention team a real workload to validate the new dashboard and alerting rules against, rather than a quiet location where problems might not surface.
Gateway Manager and Protect went in together at each site, so a store never went live with cameras but no AI filtering — avoiding a period where the dashboard would have been as noisy as the old motion-sensor DVRs.
Entrance Manager followed a few weeks behind camera rollout at each site, timed so badge issuance for existing staff could happen without disrupting daily opening and closing procedures.
The remaining ten stores rolled out over the following weeks at a steady cadence, each benefiting from alerting-threshold tuning already learned at the first two.
Challenges along the way
Twelve stores, twelve different camera vintages
Some stores had cameras recent enough to integrate directly; others had older analog units with no network path at all. Merilsoft used encoder boxes to bring the older analog cameras onto the same Gateway Manager dashboard rather than forcing a full camera replacement everywhere, keeping the rollout on budget without leaving any site with a second, disconnected system.
Tuning AI detection so it didn't just move the false-alert problem
The first week at the pilot stores still produced more alerts than the loss-prevention team wanted, mostly from delivery activity near loading doors. Detection zones and time-of-day rules were tuned store by store during the pilot so an alert reliably meant something worth a look, before the same rules were carried to the rest of the chain.
The impact
Loss prevention no longer starts an investigation with a phone call — any store's footage, live or recorded, is a lookup away on one dashboard, and AI detection means the team is reviewing actual person and vehicle events instead of sifting through motion-triggered noise from wind, weather, or wildlife.
Stockroom and back-office access moved from a physical key system nobody could fully account for to badge access logged by name and timestamp at every door, at every store. Offboarding a departed employee is now a two-minute change instead of a locksmith visit, and an incident investigation can finally answer who was actually on-site, not just who might have still had a key.
Additional value delivered
The tuning work at the pilot stores turned into something the chain kept. Detection zones and time-of-day rules worked out against real delivery activity near the loading doors at the first two locations became the starting configuration for every store after them, so later sites went live with alerting that was close to right from the start instead of beginning from scratch. What the loss-prevention team learned in that first week of false alerts got spent once, not twelve times.
Replacing the keys was the objective; the log that came with it turned into an investigation tool of its own. Every entry carries a name and a timestamp, so a question about who was in a stockroom at 2am has an answer that doesn't depend on footage existing or being clear enough to read. Access defined by role rather than by key also means head office can see exactly what any given role can open and change it, instead of trying to work out where a re-cut key might still fit.
The rollout also closed off the pattern that created the problem. Each store bought its own security equipment whenever it happened to need some, which is how 12 locations ended up with 12 disconnected DVRs and no shared view. There's now one dashboard and one access system for the chain to add to, rather than twelve independent decisions to reconcile later. And because the older analog cameras were brought in on encoder boxes rather than ripped out, equipment the chain already owned kept working — replacing those cameras became something that can happen store by store, on each store's own schedule, rather than a precondition for getting a single view at all.
Lessons learned
- Starting the rollout at the highest-incident stores gave the alerting rules a real stress test before they reached quieter locations, instead of tuning them somewhere problems wouldn't surface.
- Encoder boxes for the older analog cameras kept the project on budget without leaving any one store on a second, disconnected system during the transition.
- AI detection is only as good as its tuning — the first week of false alerts at the pilot stores was expected, and building in time to tune detection zones before wider rollout paid off across the rest of the chain.
The chain's 12 stores were never short on cameras or door locks — they were short on any of it being connected. Putting every site behind one dashboard and one access system turned twelve separate blind spots into a single, watched network.
Your operation could be next
Tell us how you run today — we'll show you what changes, with numbers.