Retail and Commercial Loss Prevention: What Actually Works

“Shrinkage” is a single line in a stock report and at least four unrelated problems underneath it: theft by customers, theft or misuse by staff and contractors, supplier and receiving errors, and administrative loss — mis-scans, wrong price files, unrecorded write-offs, damaged goods never booked out. They have almost nothing in common except that they all end up in the same number, which is why so many businesses spend on the wrong control and see the line refuse to move.

So the first task is not a security purchase. It is a two-week separation exercise: for the next fortnight, every loss that is discovered gets a category before it gets a reaction. A missing item found empty-packaged on a shelf is one thing; a delivery that was signed for but arrived two cases short is another; a product that scans at the wrong price four hundred times is a third. Almost every business that does this discovers its loss is concentrated in a category it was not spending anything on.

This matters commercially because the four categories have completely different fixes and completely different costs. Customer theft responds to layout, sightlines, and presence. Internal loss responds to process design and segregation of duties. Receiving loss responds to counting discipline at the back door. Administrative loss responds to data hygiene and costs almost nothing to fix — and it is very often the largest single component, sitting there quietly while the business installs cameras over the shop floor.

The cheapest loss-prevention control in retail is the shape of the room. High-value, small, easily-concealed stock placed near an exit and out of sight of the till is an invitation that no camera reverses. Move the same stock behind the counter, or into a sightline the staff naturally hold, and a large part of the opportunity disappears without a single purchase.

Walk the floor from the door, not from the office. Where can a person stand and be seen by nobody? Which aisle end is blind from every till position? Which fitting room, corner, or stock overspill creates a private space inside a public shop? Where does the queue form, and does it obscure the exit at the busiest hour? Most shops have two or three such spots, everyone who works there could name them, and nobody has ever been asked.

The second free control is staff behaviour, and it is not confrontation. Being greeted on entry, being approached and offered help, being noticed — these are ordinary good service and they are also the most effective deterrent in a shop, because the whole method of opportunistic theft depends on not being individually registered by anyone. A shop where every customer is acknowledged within fifteen seconds is a materially harder target than an identical shop where nobody looks up, and the difference costs nothing.

Customer theft is taken one item at a time. Receiving loss is taken by the case, and it is far less watched. The pattern is consistent across almost every business that examines it: deliveries are signed for by whoever is nearest, counted approximately or not at all, and reconciled days later by someone who was not there and cannot dispute anything. The paperwork agrees with itself and the stock does not agree with the paperwork.

The fix is procedural and cheap. Deliveries arrive in a booked window rather than whenever. One named person receives, and that person does not also enter the goods into the system — the two roles are separated because the combination is what makes an error invisible. High-value lines are counted at the door and signed for by count rather than by pallet. Discrepancies are recorded at the moment of receipt, not raised later, because a discrepancy raised later is a negotiation and a discrepancy recorded at the door is a fact.

The same door handles waste, and waste is the classic route out. Bins, cages, and returns pallets leaving the building unchecked are a mechanism, not a suspicion — which is why sensible operations locate the bin store where it can be seen, empty it on a schedule rather than ad hoc, and make removals a two-person or logged activity. None of this implies anything about the people doing the job. It removes the situation in which an honest person can be wrongly accused and a dishonest one cannot be detected.

Internal loss is the category businesses handle worst, because the instinct is to look for a person when the productive move is to look at a process. A shop where one individual can take a return, issue a refund, adjust the stock record, and void the transaction has not identified a dishonest employee — it has built a role in which dishonesty is undetectable and honesty is unprovable. That is a design fault, and it will eventually find someone.

Separation of duties is the whole answer and it costs nothing but attention. Whoever receives goods does not book them in. Whoever authorises a refund is not the person who processed the sale. Whoever adjusts a stock record is not the person who counted it. Whoever holds the safe key is not the person who reconciles the safe. In a small business full separation is impossible, and the workable substitute is a second signature and a periodic review by someone outside the daily routine.

Then use exception reporting rather than surveillance. Most point-of-sale systems will already tell you which operator issues the most refunds, the most voids, the most price overrides, the most no-sale drawer openings — ranked, per hour worked, so a busy operator is not flagged simply for being busy. The point is not to accuse the top of the list. It is that an outlier is a question, and asking it early and neutrally resolves the overwhelming majority of cases as training gaps or broken processes, which is what they usually are.

A uniformed officer in a retail environment does three things well and one thing badly. They deter at the door, which suppresses opportunistic entry before it begins. They handle the situations staff should never be asked to handle — an aggressive customer, an intoxicated person, a dispute escalating near a till. And they are the person who deals with an incident so that the shop keeps trading rather than stopping to manage a crisis with the manager's phone.

What they must not do is detain, search, or pursue on their own initiative, and this is where retail security most often goes wrong. An officer who stops the wrong person, or the right person in the wrong way, converts a small stock loss into a serious dispute in which the shop is the party at fault. The instruction has to be explicit, written into the post orders, and identical across every officer who covers the site: observe, record detail, report, and escalate — including to official services where the situation requires it.

Placement decides whether the officer is worth the money. Standing at the entrance during trading hours buys deterrence and greeting. Standing at the back door during a delivery buys receiving discipline. Being present at opening and closing buys the two moments when cash and stock are most exposed and staff are most alone. A guard positioned by habit rather than by exposure is an expense; the same guard positioned against your incident data is a control. The service structure is set out on the professional security guarding page.

Retail camera systems fail in a predictable way: wide shots of a shop floor at a resolution that shows a person was present but not who they were or what they took. A camera earns its place by answering one specific question — who handled this till, what left through this door, who was at this shelf, what did this delivery actually contain. Specify the question first and the position, angle, and resolution follow from it.

Four positions carry most of the value in a typical shop: the till, framed tightly enough to read a transaction; the receiving door, framed to see a count; the entrance, framed at face height rather than from a ceiling corner; and the highest-value fixture. Linking camera time to transaction time — so a till exception can be looked at directly rather than searched for — turns a passive recorder into an investigative tool and is frequently a software setting rather than a purchase.

Retention length is the setting that decides whether any of it works. Loss in retail is usually discovered at a stock count, which means the footage you need is from a period that has already passed — and a system that keeps a fortnight when your count cycle is monthly will be empty exactly when you look. Set retention against your discovery cycle, not against the recorder's default, and check quarterly that recording is genuinely present across the whole window. The wider trade-offs are covered in manned guarding versus CCTV.

Loss numbers are only useful if they are comparable, and most retail measurement fails on this. Total loss by store punishes the busiest branch. Loss by shift punishes whoever works Saturday. Anything measured against a stock file that has never been cleaned measures the stock file. Before drawing a conclusion, normalise by turnover or units handled, exclude the categories you have already established are administrative, and confirm that the underlying data is not itself the problem.

Then act on categories rather than on people, at least first. If receiving loss is the outlier, the intervention is counting discipline at the door and a booked delivery window, not a conversation with a storeman. If till exceptions cluster on one operator, the first move is to check whether that operator was trained differently or works the returns counter, because both explain the pattern innocently. Reserve the individual conversation for after the process explanations have been eliminated — it is fairer, and it is also more accurate.

Finally, review on a fixed rhythm and change one thing at a time. A business that alters layout, adds a guard, tightens receiving, and changes its refund policy in the same month will never know which of the four worked, and will keep paying for all of them. Change one, measure for a full count cycle, keep it or drop it, then change the next. That discipline is slower and it is the only way the spend becomes evidence rather than faith — the same principle applied to guarding in supervision and accountability.

Put the whole thing in order and the sequence is unglamorous. First, separate the four categories for a fortnight so you know which problem you have. Second, clean the administrative loss, because it is usually significant and almost free to fix. Third, fix the back door: booked deliveries, a named receiver, counting at the point of receipt, and controlled waste removal.

Fourth, redesign the room: move the high-value, easily-concealed stock into a held sightline, close the blind corner, and set the expectation that every customer is greeted. Fifth, separate the duties around refunds, voids, stock adjustments, and cash, and turn on exception reporting. Only then, sixth, buy presence and coverage — an officer placed against your actual incident pattern, and cameras framed to answer specific questions with retention matched to your count cycle.

Done in that order, most businesses spend less than they expected and see the number move, because the expensive controls are aimed at a problem that has already been narrowed. Done in the reverse order — cameras first, then a guard, then eventually a look at the data — most businesses spend more and see the number hold, and conclude that loss prevention does not work. If your site needs the exposure mapped before the spend, that is a risk assessment; if the answer is presence at specific hours, that is access control and electronic security.