Use case · Retail shrink
One number at the end of the count, three different causes
External theft, internal theft and process error arrive as a single figure and get treated as a single problem. A camera system addresses part of one of them well, part of another indirectly and none of the third. This page is about which part, and what the sourced data actually supports.
- CSLB C-7 #976737
- BSIS ACO #7501
- CMMC certified
- SAM.gov registered
- Est. 2008
The problem
You cannot fix a number you have not split
Shrink turns up as a variance: the system says you hold this many and the count says you hold fewer, and the difference gets a name and a line in a report. What the number does not tell you is which of at least three unrelated things produced it.
- Somebody walked out with goods.
- Somebody who works for you moved goods or money.
- Nobody took anything at all, and the discrepancy is a receiving error, a mis-scan, a markdown never entered, a transfer never recorded or a damage write-off that never happened on paper.
Those three have nothing in common operationally. The first is about the sales floor and the exits. The second is about the back of house and the till. The third is about process and is not a security problem at all — no camera ever installed has corrected a paperwork error. Yet because they arrive together, the response is usually a single undifferentiated one: more cameras, everywhere, and a hope that the number moves.
What that costs is not just the equipment. It is the store manager who now spends two hours a week scrubbing footage with no idea what they are looking for. It is the staff who correctly notice they are being watched all day and correctly resent it, on a shrink problem that turned out to be receiving. It is the district manager comparing stores on a figure that means something different in each of them. And it is the genuine external-theft problem in the one store that has it, buried inside a company-wide average.
It resists solving for one more reason, and it is the reason most retail camera systems disappoint: the useful coverage is not of space, it is of transitions. Shrink happens where stock changes hands or changes status — at the counter, at the threshold, at the stockroom door, at receiving, at the waste stream, at returns. A system that covers the floor evenly at a resolution that shows movement is covering all the places nothing happens.
What the data says
Retail theft is down, and the argument is elsewhere
The National Retail Federation's most recent study finds shoplifting falling, and we are going to lead with that, not bury it. The finding that actually matters commercially is a different one from the same report.
Shoplifting incidents down 12.4%
NRF's The Impact of Theft & Violence 2026, covering fiscal 2025, reports a 12.4% decrease in shoplifting incidents and an 8.1% decline in retail merchandise theft against 2024. ESTIMATE from a voluntary industry survey.
63% report fewer than half of incidents
From the same study: 63% of retailers report fewer than half of their incidents to law enforcement. For that share of events no external record is created at all. This is the genuinely useful finding on this page.
The last published shrink rate is FY2022
The most recent shrink rate NRF has published was 1.6% of sales for fiscal 2022, $112.1 billion, up from 1.4% and $93.9 billion in fiscal 2021, with internal and external theft accounting for nearly two thirds — 65% — of it. NRF's newer reports no longer publish a shrink rate at all.
Sources. Fiscal 2025 figures: National Retail Federation, The Impact of Theft & Violence 2026, released 30 July 2026 — nrf.com. Type: ESTIMATE from a voluntary, self-selected survey of 66 retail companies representing 143 brands and $1.7 trillion in fiscal 2025 sales, about 31.6% of total US retail sales. Shrink rate: National Retail Federation, 2023 National Retail Security Survey — nrf.com. Period: fiscal year 2022, roughly four years old at the time of writing. Type: ESTIMATE from a voluntary survey of 177 retail brands, approximately $1.6 trillion in annual sales across more than 97,000 US locations. Retrieved: 2 September 2026. These are two different reports and their numbers are not directly comparable — the sample fell from 177 brands to 66 companies between them, so do not read a trend across the two. Neither is a census and neither is government data.
The direction is down, and any page implying a national retail theft surge is contradicting the most recent primary source. That is not a comfortable thing for a security company to write and it is what the data says. The commercially useful finding is not volume, it is the 63%: most incidents never reach law enforcement, which means for most of them the only account that will ever exist is the one your own system captured. That raises the standard your system has to meet.
| Year | Shoplifting | Theft from building | All larceny / theft |
|---|---|---|---|
| 2020 | 1,272 | 616 | 16,124 |
| 2021 | 1,179 | 518 | 18,113 |
| 2022 | 1,181 | 427 | 15,746 |
| 2023 | 1,423 | 440 | 16,144 |
| 2024 | 3,097 | 387 | 17,133 |
| 2025 | 4,221 | 370 | 16,399 |
Source: San Diego Police Department, Police NIBRS Crime Offenses, City of San Diego Open Data Portal — data.sandiego.gov/datasets/police-nibrs. Period: calendar years 2020–2025. Type: COUNT of offences reported to police. Retrieved: 2 September 2026. Read this carefully. The number of shoplifting offences recorded by SDPD rose from 1,181 in 2022 to 4,221 in 2025, while total larceny and theft over the same period was flat to slightly down. A category rising several times over while its parent category does not move is the signature of a classification or recording change, not necessarily a change in underlying behaviour. The dataset carries no explanation for the discontinuity and we could not verify one, so we are not going to describe this as a surge, attribute it to any law, policy or enforcement change, or use it as an argument for anything. It is presented because it is in the data and omitting it would be selective.
The full statistical treatment — every NRF figure, the San Diego series and the methodology in detail — is on our retail shrink statistics page. This page is about what to do about the problem.
Why it resists solving
Coverage of space, when the problem is transitions
The default retail install is a camera count derived from square footage, mounted high and wide because that covers the most floor per unit, at a resolution set to whatever keeps the storage bill down. Every one of those decisions is reasonable in isolation and together they produce a system that reliably shows that something happened and reliably cannot show who.
The alternative is to design against transitions. There are only a handful of points in a store where stock changes hands or changes status: the register, the entrance, the fitting room approach, the stockroom door, receiving, the waste stream, the safe or cash office, and whichever displays hold the goods worth the risk. Cover those eight properly — framed to identify, specified for the light that is actually there — and you have a system that answers questions. Cover the whole floor adequately and you have footage of a store.
The second obstacle is review. A store generates far more footage than anybody will ever watch, so a system that can only be searched by scrubbing a timeline is a system nobody uses after the first month. What makes review sustainable is being told when to look: a report of voids, refunds, no-sales and manual discounts, an access event on the stockroom door, a delivery record. All of which depend on the clocks agreeing, which is a commissioning step that gets skipped more often than any other.
And the third is that a large part of shrink is not a security problem and no amount of security spending will move it. If your variance is receiving error, the answer is a check against the manifest at the door, not a camera above it. We would rather tell you that during a survey than take the order.
What reduces it
Six things that change the outcome
Aimed at the part of shrink a system can genuinely address. We will not attach a reduction figure to any of it, because we cannot evidence one for your store.
The transition positions, done properly
Register, entrance, fitting room approach, stockroom door, receiving and high-value displays — each framed for what it has to prove, not for how much floor it covers. That is commercial security camera installation.
Credentials on the back of house
Stockroom, receiving, cash office and any room holding customer or personnel data, on a credential instead of a key — which is what turns a recorded event into an attributable one. That is commercial access control.
Coverage after close
For stores exposed to forced entry or smash-and-grab, an operator watching during closed hours able to speak into the space, which interrupts instead of records. That is remote video monitoring.
Intrusion and duress at the boundary
Monitored detection on the doors and the shutters, plus duress provision where staff open or close alone — a different requirement from shrink and one that turns up in the same conversation. That is commercial alarm monitoring.
Cabling that survives a fit-out
Leased units bring landlord consent for routes, penetrations and ceiling access, and existing cabling of unknown provenance. Getting this right is what allows a position to be added later. That is structured cabling.
A system that is still doing its job
Lenses collect dust, cameras get knocked by stock handling, drives fail quietly and retention drifts below what a count six weeks later needs. Discovered when the footage is needed. That is CCTV maintenance and repair.
Who has it worst
Where stock goes missing at scale
Shrink is a retail word but the problem is not only retail's. Three types carry it in different forms.
- Retail. The full version: public trading space, transactions, returns, a stockroom, a receiving door and staff handling goods all day, with the added complication that the coverage has to be designed around real privacy boundaries. That is retail security and loss prevention.
- Warehouses and distribution. Same problem at pallet scale, where a discrepancy is discovered at a cycle count weeks later and the transitions are dock doors, pick faces, staging and the yard, not a counter. That is warehouse and logistics security.
- Managed and multi-tenant property. Loading bays, goods-in areas, package and mail rooms and shared service corridors, where the loss belongs to a tenant and the coverage belongs to the landlord. That is office and property management security.
The other verticals are on all industries and the full service list is on all commercial services. The related problems are theft and burglary for forced entry after close and employee safety where the concern has moved from stock to people.
One line we hold on every retail site, stated here because a retailer may land on this page first. Cameras cover the entrance to a fitting room area and the approach to it, and never the interior of a fitting room, a restroom or a staff changing area. That is not a configuration option or something to be negotiated per site; it is a line we do not cross, and we will decline the request if it is made. And the scope boundaries: we do not supply loss-prevention officers, which is licensed guard work coordinated through Guardian Integrated Security, Inc., of which we are a DBA; we do not hold a C-10 or a C-16, so we do not work on fire alarm systems, which is coordinated with your own licensed fire alarm contractor; and we work on commercial premises only.
At survey
What we would look at in your store
A shrink survey starts with your numbers, not with the building, because the building cannot tell us which of the three problems you have.
How the variance breaks down
By category, by area and by store where there is more than one. External, internal and process error look completely different in a breakdown and identical in a total, and this is the step that decides whether we should be quoting anything at all.
The transitions, walked
Register, entrance, fitting room approach, stockroom door, receiving, waste stream, cash office and high-value displays. What each currently sees, at what framing, and whether it would identify a person or merely record one.
The waste stream
Cardboard, cages, the compactor and where they sit relative to the back door. It is the quietest route out of a store and it is almost never covered, because it does not look like a door.
What your till system exposes
Whether the point-of-sale can produce an exception report, in what form, and whether anything else can consume it. We check this at survey and tell you plainly what your system supports rather than naming a product as supported on the strength of a datasheet.
Clocks, retention and who can review
Whether the recorder agrees with the till and the door controller, what is genuinely being retained against how long a count takes to surface a problem, and which named people can actually pull footage.
The lease and the fit-out
Routes, penetrations, ceiling access and make-good at lease end all need landlord agreement, and older units carry cabling nobody can account for. Established before work is scheduled, not after somebody objects.
Findings go in writing whether or not you go on to contract with us, including the positions we think are not worth a camera and the parts of your shrink we do not think we can help with. On this problem that second list is often the more useful one.
Common questions
Retail shrink questions
Not with cameras. Start by splitting the number, because shrink is at least three unrelated problems reported as one figure — external theft, internal theft, and process and paperwork error — and they have almost nothing in common except the line they land on. A count variance caused by a receiving error is not addressed by anything a security contractor sells. Cycle counts by category and by area, a look at where variance concentrates, and a comparison of like stores will usually tell you which of the three you actually have before anybody spends anything. Once you know that, the security question becomes narrow and answerable: which specific transitions need to be visible and reviewable. We would rather have that conversation than quote a camera count against your square footage.
RelatedRetail shrink statistics
There are fewer routes than the anxiety suggests, and listing them is the most useful hour of a survey. The front door, carried or concealed. The fitting room area, in and out. The receiving door, during a delivery when it is open and busy and nobody is watching it. The waste stream — cardboard, cages, compactor — which is the quiet one. The staff exit and the staff room. The counter itself, through a transaction that was never a real transaction. And returns, processed against goods that were never bought. Every one of those is a place, and a place can be covered and reviewed. A sales floor cannot be, usefully, which is why blanket floor coverage disappoints so consistently.
RelatedRetail shrink statistics
By searching a list, not a timeline. The practical arrangement is that some other system tells you when to look — a report of voids, no-sales, refunds, manual discounts and returns, an access event on the stockroom door, a delivery record — and the footage is retrieved at those timestamps. That turns an eight-hour scrub into a few dozen ten-second looks. It depends absolutely on the clocks agreeing, which is a commissioning detail people skip and then wonder why nothing lines up. Where a platform can bring the two together on one screen it is better again, but the manual version works perfectly well and costs nothing beyond getting the time right.
RelatedRetail shrink statistics
Not on its own, and anybody who says otherwise is selling you something. A camera records what happened; it does not attribute intent or employment status. What produces attribution is the combination of a recorded event and a second record that names somebody — a credential on the stockroom or receiving door, an operator login on a till, a delivery signature. That is why access control matters more to an internal-shrink question than another camera does. It also changes the tone of the exercise in a way that is worth noticing: reviewing exceptions against records is a narrow, evidence-led activity, whereas watching staff work all day is neither pleasant nor productive.
RelatedRetail shrink statistics
Receiving is the most under-covered position in most stores and the one with the clearest path from stockroom to street. Covering it properly means a view of the door from both sides, so you can see what came off the vehicle and what went back on it, and a record of who opened the door and when — a credential, not a key. The door being propped during a delivery is normal and unavoidable, and an alert when it stays open beyond a set period is a cheap thing to add and disproportionately useful. Beyond that it is process: a receiving check against the manifest at the point of delivery, not an hour later, which is not something we sell but is frequently the actual answer.
RelatedRetail shrink statistics
It changes where the visibility has to be, not how much of it there is. A self-checkout area concentrates a lot of individual transactions in a small space with no staff member present for most of them, so the useful position is one that shows the item, the scanner and the person together at a framing that supports a review — not a wide shot of six terminals from the ceiling. The same exception-review logic applies and applies more strongly, because the system generates its own detailed record of what was scanned and what was not. Whether your particular terminals expose that record in a form anything else can consume is a question we check at survey and answer plainly, not promising an integration before seeing what you run.
RelatedRetail shrink statistics
They are genuinely useful and they are mostly useful for something else. Counting, dwell and flow tell you about staffing, layout and conversion, and a store that uses them for merchandising is getting real value. For shrink specifically they are indirect: knowing that an aisle is busy at four does not tell you anything went missing from it. The analytics that touch shrink directly are narrower — a line crossed at a receiving door outside delivery hours, a door held open, presence in the stockroom when nobody is rostered — because each of those produces a specific event you can go and look at. We would rather scope those than sell a feature list, and we do not name a platform as supporting something on the strength of a datasheet.
RelatedRetail shrink statistics
That is a commercial decision for you and there may be insurer or brand requirements that settle it. What we can tell you is what the most recent primary data says about the practice: the National Retail Federation reports that 63% of retailers report fewer than half of their incidents to law enforcement, from its 2026 study of 66 retail companies. That is a voluntary, self-selected survey, not a census. The consequence, whichever way you decide, is the same and it is the reason this page exists: for most incidents no external record is created, so the only account of what happened is the one your own system holds. That raises the standard your system has to meet rather than lowering it.
RelatedRetail shrink statistics
No loss-prevention officers, no store detectives. That is guard work under a Private Patrol Operator licence — PPO #121089, which belongs to Guardian Integrated Security, Inc. San Diego CCTV Pros is a DBA of Guardian, so where a store genuinely needs a person it is arranged through them under that licence and never sold from here. The systems are ours: CSLB #976737 for cameras and cabling, BSIS ACO #7501 for monitored CCTV and alarm.
RelatedRetail shrink statistics