Retail Loss Prevention: What Works and What Doesn't
Retail loss prevention is the work of finding where a shop loses stock and cash, then cutting each loss: theft by customers, theft by staff, errors and supplier losses. UK retailers counted 5.45 million customer thefts costing £408 million in the year to August 2025. Start by counting your own losses. Cameras and tags come after that.
We build AI that watches CCTV, so we have an interest in cameras. This guide still tells you where the evidence for them is thin.
Where to start:
- You do not know where the stock is going: read where the losses come from, then how to measure yours.
- You are about to buy equipment: check what the evidence says works first.
- You have a thief on camera: go to what the police will do.
What is retail loss prevention?
It is everything a retailer does to keep the stock and cash it should have. The number it works on is shrinkage, also called shrink: the gap between the stock your records say you hold and the stock you find when you count.
The term covers more than catching thieves. Loss prevention includes how deliveries are checked, who can approve a refund, how the shop is laid out and what staff are told to do when someone walks out without paying. In a chain it is a department. In a single shop it is the owner, a stock count and a few rules.
Researchers split loss two ways. Professor Adrian Beck's Total Retail Loss work, published by the ECR Retail Loss Group, separates known loss, where you can say what happened, from unknown loss, where stock is simply missing. It then separates malicious loss (someone meant to take it) from non-malicious loss (a mistake, damage, waste). His advice is not to guess the causes of unknown loss. Most shops do guess, and they guess "shoplifters".
Where the losses come from
Customer theft is the largest single cause and still only about a third of the total. The last full UK breakdown is the Centre for Retail Research's 2019 study of 264 retailers with 19,000 shops, which put average shrinkage at 1.42% of sales.
| Cause | Share of shrinkage | What it looks like |
|---|---|---|
| Theft by customers | 34.6% | Shoplifting, organised gangs, walk-outs at self-checkout |
| Errors | 25.3% | Wrong prices on the till, wrong change, deliveries booked in wrong |
| Theft by staff | 22.1% | Fake refunds, items not scanned for friends, stock left by the bins |
| Suppliers and warehouse | 18.0% | Short deliveries, fraud and pilfering before stock reaches the shelf |
These are retailers' own estimates and they are from 2019, before the rise in shop theft, so read them as proportions and not as today's totals. Two things hold up. Theft by customers and staff together is 56.7% of the loss, which leaves over 40% that no camera or guard will touch. And a quarter of it is error, which costs nothing but attention to fix.
The current numbers on crime come from the British Retail Consortium's 2026 crime report: 5.45 million known customer thefts in the year to August 2025, costing £408 million, which is 70% of the £583 million retailers lost to known theft and damage. The BRC changed its method that year to count only known loss, so these figures cannot be compared with the £2.2 billion in its 2025 report.
US readers will know the National Retail Federation's figure of 1.6% average shrink for 2022. The NRF stopped publishing its shrink survey in 2024, so that is the last one.
What one theft really costs
A stolen item costs you far more in sales than you paid for it. You paid for it, and the only way to get that money back is out of profit on other sales. The thinner your margin, the more you have to sell. Drag the slider to your own net margin: the profit left from each £1 of sales after stock, wages and rent.
The lower a shop's profit margin, the more it must sell to pay for one theft. A stolen bottle that cost the shop £10 takes £333 of sales to earn back at a 3% margin, £1,000 at 1% and £67 at 15%.
- Stock stolen
- Sales needed to pay for it, £10 a basket
At a 3% margin, a stolen bottle that cost you £10 takes £333 of sales to earn back. UK supermarkets averaged an operating margin of 3.2% in 2021/22 and 1.8% the year after, according to the Competition and Markets Authority's 2023 groceries review. At 1.8% the same bottle takes £556 of sales.
The same sum tells you what prevention is worth. A shop with £500,000 of sales and average shrinkage loses about £7,100 a year, or £137 a week. Anything that costs more than the loss it removes is a bad purchase, however good the demonstration was.
How to measure your losses
You cannot cut a loss you have not located. Four habits find it.
- Work out your shrink rate. Stock you should have, minus stock you count, divided by sales for the period. If the records say £42,000 of stock, the count finds £40,200 and you sold £125,000 that quarter, shrinkage is £1,800 and the rate is 1.44%.
- Count by line, not just in total. A yearly count gives you one number and no clue. Count your 20 most stolen or most valuable lines every week. A loss that shows up in spirits and razor blades is theft. A loss spread evenly across the shop is more likely process.
- Read the till's exception reports. Refunds, voids, no-sale drawer openings and price overrides, listed by member of staff. One name far above the others is worth a look at the till camera.
- Keep an incident log. Date, time, what was taken, value, what the person looked like, which clip shows it, whether you reported it. Three months of it shows you the hour, the door and often the handful of people responsible.
Expect the records themselves to be wrong. An ECR study of about a million product lines across seven European retailers found about 60% of stock records did not match the shelf, and that correcting them raised sales by 4% to 8%, because the system stopped believing in stock that was not there. A count pays for itself before it catches anyone.
What the evidence says works
Less than the brochures say. We looked for independent evidence behind each common measure.
| Measure | Independent evidence, and our reading |
|---|---|
| Security tags | A review of eight studies found no statistically significant effect overall.
Five showed less theft or shrinkage, one no change, two more. Visible tags
did better than discreet ones (College of Policing) Our reading: Worth it on the lines you lose most, with tags people can see. |
| CCTV | Across 76 studies, crime fell 13% in places with CCTV. Cameras alone had no
significant effect; cameras with several other measures, 34%. No effect on
violence. Shops were not studied separately (College of Policing) Our reading: Buy it for evidence. Do not expect it to deter by itself. |
| Greeting customers | Police advice, no study we could find. The Metropolitan Police says greeting customers "can put off shoplifters" Our reading: Free, so do it. |
| Screens showing customers the camera view | One US trial in 47 shops reported a positive effect. A later ECR review calls the research "limited" Our reading: Cheap to try at the door or the self-checkout. |
| Self-checkout | Raises loss. See below Our reading: Budget for the loss and supervise it. |
| Telling staff what you check | One grocery retailer that told its shops a new remote CCTV team was watching
them reported shrinkage 15% lower than in matched shops over two years (ECR) Our reading: The cheapest measure against staff theft. |
| AI that spots concealment | Suppliers' and retailers' own claims only Our reading: Ask for a trial in your shop before you pay. |
The CCTV row is the one to remember. A camera works when something follows from it: a member of staff who walks over, a report the police act on, a ban that is enforced. The review points the same way: schemes someone watched live cut crime 15%, cameras alongside several other measures such as lighting and signs cut it 34%, and cameras alone showed no significant effect. A dummy camera is a poor saving for the same reason: nothing can follow from it, and it gives the police no footage.
Cutting customer theft
Start with what costs nothing. The trade body for convenience stores, the ACS, reports in its 2026 crime report that retailers say "a small number of repeat offenders are responsible for the majority of shop theft offences", driven by addiction, opportunity and organised crime. The items taken most are alcohol, confectionery and meat. So the job is narrower than it feels: a few people, a few shelves.
- Move the target stock. Spirits, meat, razor blades and baby formula go where the till can see them, or behind it. Put out three of a costly line, not thirty.
- Clear the sight lines. Low shelving in front of the counter and no display blocking the view of the door.
- Greet everyone who comes in. A thief does not want to be noticed.
- Know your regular thieves. The incident log gives you faces and times. Ban them in writing and join your local business crime partnership, which shares who is banned between shops.
- Tag the lines the count says you lose. Not the whole shop.
- Fix the door camera. It has to show a face well enough for a stranger to be identified, and it is often mounted too far away. Our guide to CCTV for retail shops has the distances.
What staff should watch for
Behaviour, never appearance. The common signs:
- Watching the staff or the cameras more than the shelves.
- Picking up and putting back the same costly item several times.
- An empty or open bag, a pushchair hood, or a coat on a warm day.
- One person keeping staff talking while another moves around the shop.
- Going into a fitting room with five items and coming out with three.
- Heading for the door as soon as a member of staff comes near.
A sign is a reason to walk over and offer help. It is not a reason to accuse anyone. A thief who thinks they have been spotted is more likely to leave.
Refund fraud
Stolen goods are often turned into cash at the refund desk of the shop they came from. Three rules close most of it:
- No cash refund on unwanted goods without a receipt. Offer an exchange or a credit note. Faulty goods are different: the customer needs only proof of purchase, such as a bank statement.
- Refunds go back to the card that paid.
- Log every refund without a receipt, with a name. The same person three times in a month is a pattern.
For the equipment itself, from tags to alarms to panic buttons, see our guide to retail security systems and which loss each stops.
Self-checkout theft and losses
Self-checkout saves wages and raises shrinkage, and the second part is rarely in the sales pitch. ECR's 2026 self-checkout loss report, with data from 39 retailers, found losses 33% higher in stores with self-checkout than in stores without, and 22% higher on average in the year after the machines went in.
Not all of it is theft. The same report found retailers' estimates of how much non-scanning is deliberate ran from 6% to 80%, and calls their methods unreliable. People mis-scan, pick the wrong loose item and give up on a barcode that will not read. What helps:
- A member of staff standing at the machines, not watching from a till.
- One way in and one way out of the self-checkout area.
- Costly lines sold only at a staffed till.
- A screen above each machine showing the shopper their own camera picture.
- Scales in the bagging area that check each item's weight. They catch unscanned items and also stop honest shoppers, so someone has to be there to clear the alerts.
Cameras that check each item against what was scanned were on trial in 2025 at UK chains including Home Bargains and Tesco, according to The Grocer. We have seen no independent measurement of how well they work.
Cutting staff theft
Staff theft was 22.1% of shrinkage in the 2019 study, and it is the loss owners least like to look for. Research by Professor Emmeline Taylor published by ECR estimates that retailers "probably detect and close just 2% of employee theft cases".
The common methods are few:
- Fake refunds on goods that were never returned.
- Sweethearting: not scanning items, or scanning a cheaper one, for a friend or relative.
- Voids and discounts applied after the customer has paid cash.
- Stock out of the back door, often left by the bins for later.
What cuts it:
- Tell staff what you check. In the same research, a grocery retailer that told its shops a new remote CCTV team was watching reported shrinkage 15% lower than in matched shops over two years. People steal less when they believe someone is looking.
- A login per person on the till, never a shared one.
- Two people for refunds over a set value, and a weekly look at the exception report.
- A camera over the till and one on the back door. Staff must be told about every camera that can see them. Our guide to CCTV law in the workplace has the rules, including why hidden cameras are almost never lawful.
- Bag checks and bin runs done by two people, written into the contract. Even with the clause you need the person's consent each time. Never search by force.
- Cash counted by two people. Lift large notes from the till into a safe through the day, and have a different person check the day's takings against the till report.
Cutting errors and supplier losses
Over 40% of shrinkage in the table above has no thief behind it, or none in your shop. The fixes are dull and cheap.
- Check deliveries against the delivery note before signing, at least for the lines that matter. A short delivery signed for is stock you never had.
- Check shelf prices against the till whenever prices change.
- Record waste when it happens. Damaged, out of date and marked down stock that is not written off looks exactly like theft at the next count.
- Keep the stock room shut, and the back door shut during deliveries.
Do these before buying any equipment. If the count is wrong you cannot tell whether the equipment worked.
The law: theft, the £200 rule and stopping people
This section covers England and Wales. Scotland and Northern Ireland have their own law and courts, and the £200 rule below does not apply in either. Shoplifting is not a separate offence. It is theft under section 1 of the Theft Act 1968, with a maximum of seven years in prison.
The "£200 rule" is real and widely misread. Since 2014, theft from a shop of goods worth £200 or less is tried in a magistrates' court, where the maximum sentence is six months, unless the defendant chooses a Crown Court trial. It is still a crime, and the police can still arrest and charge for it.
That rule is on its way out. The Crime and Policing Act 2026 became law on 29 April 2026. Section 47 removes the £200 rule, and section 45 creates a new offence of assaulting a retail worker at work. Neither had been brought into force when we checked on 22 November 2026, and the government has given no date. Until it does, the £200 rule applies.
Can staff stop a thief?
The law allows it in narrow cases, and we would tell staff not to. Any member of the public may arrest someone committing theft, but only when a police officer cannot practicably do it and the arrest is needed to stop the person making off or causing injury or loss. Get it wrong and your member of staff is the one accused of assault.
Safety is the stronger reason. The Health and Safety Executive's guidance on violence to retail staff says training should cover "not to resist or follow violent offenders". The ACS tells shop staff to "keep a safe distance when engaging with suspected shop thieves". Write the policy down: staff watch, describe and report. They do not chase or block the door.
The risk is real. The BRC counted more than 1,600 incidents of violence and abuse against shop workers a day in the year to August 2025, down from 2,000 the year before. Three things to settle before the day it happens:
- A code phrase staff can say over the till or a headset to call a colleague over.
- Where the panic button is, and that it is for threats, not for theft. Our guide to shop security systems explains the police rules for it.
- Who calls 999, and who stays with the member of staff afterwards.
Will the police come?
Sometimes. In the BRC report, retailers reported only 15% of incidents to the police, officers attended 31% of those, and 50% of retailers rated the police response as poor, down from 61% a year earlier.
The police set out when they will attend in the Retail Crime Action Plan of October 2023. Attendance is prioritised in three cases:
- Violence has been used.
- An offender has been detained, for example by store security.
- Evidence needs securing promptly and only police can do it in person.
The plan also commits forces to recover clear CCTV footage and to check usable images against the Police National Database with facial recognition. So a sharp picture of a face is what turns a report into a name.
The official figures for England and Wales:
- The police recorded 519,516 shoplifting offences in the year to June 2026, 2% fewer than the year before (Office for National Statistics).
- 23.2% of shoplifting offences ended in a charge or summons in the year to March 2026, up from 18.5% (Home Office).
- 52.7% were closed with no suspect identified.
Compare the recorded offences with the BRC's 5.45 million known thefts across the UK. The two are counted differently, but the gap is plain: most shop theft never becomes a crime report.
How to report so it gets used
- Report every theft, online if the thief has gone. Unreported theft does not exist when police decide where to put officers.
- Send a clip, not a description. The face at the door, the theft itself, and the time on screen.
- Say when it is the same person. Linked offences are what get a repeat offender charged or banned by a court.
- Call 999 if anyone is threatened.
Organised theft is handled nationally. The police unit funded by the Home Office and by retailers through the Pegasus partnership reported 257 arrests in its first two years. That work runs on reports from shops.
What cameras and AI add
A camera records. AI is software that reads the video as it arrives, so the camera can raise an alert while something can still be done. Our post on the four levels of AI security cameras explains how far that goes. What we would trust in a shop today:
- A person in the shop or yard after closing. Built into many current cameras.
- Someone in the stock room or behind the counter who should not be there.
- Finding footage fast. Searching a day of video for a red coat takes seconds. Our guide to CCTV analytics explains each kind.
- Matching till events to video, so every refund has a clip beside it.
What to be careful with:
- Concealment detection, software that claims to spot someone pocketing stock. We found no independent test of its accuracy. A busy aisle is hard for software, and every false alert sends a member of staff to accuse a customer.
- Facial recognition. 22% of retailers in the ACS survey use it. It brings the strictest data protection duties a shop will meet and a real rate of wrong matches. Read our guide to facial recognition CCTV, its error rates and the law before a salesperson visits.
Any camera that records people brings duties under UK GDPR. Our CCTV and GDPR guide covers signs, retention and sharing images of suspects.
Our own work is on the behaviour side. For a UK CCTV security company we built real-time trespass and violence detection that runs on a computer on site, so footage stays in the building, with one mid-range graphics card handling 15 to 20 cameras. If you have cameras and one specific loss they are not catching, our computer vision service starts with a feasibility study at £4,000 to £6,000 ($5,360 to $8,040) on your own footage, credited against the build if you go ahead. The AI CCTV guide covers the cheaper options to try first.
How much to spend
Less than the loss it removes. Retailers spent £985 million on crime prevention in a year when their known theft and damage losses were £583 million, and nobody can say how much of that spend paid for itself. Convenience stores spent an average of £6,213 ($8,325) a store.
A guard shows the sum most clearly. At the National Living Wage of £12.71 an hour, one guard for 40 hours costs £508 ($680) a week in pay alone. The shop in our example loses £137 a week to everything. A guard there costs almost four times the whole loss.
The published UK prices we could find for the common purchases:
| Purchase | Price |
|---|---|
| Security tags | From about 1p each for disposable tags, 20p to 35p for reusable ones (College of Policing, 2017 figures) |
| Branded anti-theft stickers | 2p each (Southern Co-op, in the ACS report) |
| Basic four-camera CCTV, fitted | About £800 plus VAT ($1,070), Checkatrade, November 2026. See retail CCTV costs |
| Monitored alarm for a small shop, fitted | £800 to £1,500 ($1,070 to $2,010), one installer's published prices. See security system costs |
| Guard, 40 hours a week | At least £508 a week in wages ($680) |
We found no published UK price for the gates that read the tags. Suppliers quote per doorway, so get two quotes and ask for the price of the tags as well as the gate.
Violence changes the sum. Staff safety is a legal duty and not a line in the shrink figures, so a panic button or a body camera can be right even where the theft numbers would not pay for it.
What to do first
- Count. Your shrink rate, then your top 20 lines weekly.
- Fix the process losses. Deliveries, prices, waste, the back door.
- Write the staff rules. No chasing. Own till logins. Two people for big refunds. Say what you check.
- Move stock and clear sight lines.
- Log and report every theft, and ban the repeat offenders in writing.
- Get one camera right: a face at the door.
- Tag the lines the count names.
- Then consider alarms, AI or a guard, each against the loss it would remove.
Count again after three months. If the rate has not moved, the loss is somewhere else.
Questions
What are five methods of loss prevention in retail?
Counting stock by line, checking deliveries and prices, controlling refunds and till logins, keeping costly stock in sight of the till, and a door camera good enough to identify a stranger. Tags and guards come after those.
What is a good shrink rate?
The UK average in the last full study was 1.42% of sales, and the last US figure was 1.6%. Below that is good, but compare like with like: grocery and fashion run higher than furniture. A rate that falls from one count to the next matters more than the average.
Is shoplifting a crime in the UK?
Yes. It is theft under the Theft Act 1968. Theft from a shop of goods worth £200 or less is normally tried in a magistrates' court with a maximum of six months, and a law removing that limit was passed in April 2026 but was not yet in force in November 2026.
What is the £200 rule on shoplifting?
In England and Wales, theft from a shop of goods worth £200 or less is normally dealt with in a magistrates' court. It does not mean the police ignore it. The Crime and Policing Act 2026 removes the rule, but that part was not yet in force in November 2026.
Do shops keep track of shoplifters?
Many do. Shops keep incident logs and CCTV clips, issue written bans, and share details of offenders through local business crime partnerships. The Information Commissioner's Office says sharing images to prevent crime is lawful "as long as it's necessary and proportionate". Some shops use facial recognition to do it automatically.
What is the most effective deterrent to shoplifting?
No single one has strong evidence. The research favours combinations: cameras with staff who respond, visible tags on the lines that are stolen, and attention from staff at the door. Cameras by themselves showed no significant effect.
What is the most shoplifted item?
In UK convenience stores the ACS ranks alcohol first, then confectionery and meat. In any shop it is whatever is small, costly and easy to resell. Your own count will tell you.
Do most shoplifters get away with it?
Most thefts are never reported, so yes. Of those the police in England and Wales did record in the year to March 2026, 23.2% ended in a charge or summons and 52.7% were closed with no suspect identified.
What does a loss prevention officer do?
In a large retailer, they investigate losses: reading till and stock data, reviewing CCTV, interviewing staff, training shop teams and working with the police. A store detective, who watches for thieves on the shop floor, is one part of that job.