Is it cheaper to prevent crime than to prosecute it?

In almost every property-crime case, yes. The victim's loss is already spent when the investigation starts, and the public cost of processing the case is usually larger than the loss itself: about $4,127 of criminal-justice cost for a burglary where the victim lost $1,362, and about $2,879 for a larceny worth $480, in 2008 dollars. On top of that the business pays its own hidden bill in staff hours, court days, downtime, insurance and replacement. A measure that stops the incident avoids the whole chain.

What percentage of property crimes are actually solved?

In 2024 US law enforcement cleared 15.9% of reported property crimes by arrest or exceptional means, against 43.8% of violent crimes, according to the FBI. By offense the figures are about 15% for burglary, 17% for larceny-theft and 9% for motor vehicle theft. Because only about three in ten property crimes are reported to police in the first place, roughly five of every hundred property crimes end in an arrest, and a clearance is not a conviction.

Does video surveillance deter crime?

It can, under conditions. The Campbell Collaboration's systematic reviews of CCTV found a modest but significant overall reduction, with the largest and most consistent effects in car parks, a 51% drop in the 2008 review. Actively monitored systems produced significant reductions while passive, record-only systems did not, and schemes that paired cameras with other measures such as lighting did better than cameras alone. A camera that only records evidence deters less every year as people learn that recording rarely leads anywhere.

What is the see it, get it, fear it model of deterrence?

A framework from Dr. Read Hayes and the Loss Prevention Research Council (LPRC) for designing deterrents from the offender's point of view. The offender has to see the measure, get what it will do to their plan, and fear it as a credible consequence. A deterrent that fails any one of the three has no effect. Steve Lindsey uses it on the TalkLP deterrence episode to explain why visible, responsive, consequence-backed systems keep working while passive cameras fade.

How do you measure crime prevention if nothing happened?

The same way you measure any intervention: establish a baseline incident rate for the site over a period you trust, date the deployment, then count the same way afterwards. Count incidents, not arrests, and include near-misses and trespass events the system logged. Lindsey suggests looking for a reduction in the order of 70% from a well-designed deterrence deployment. Watch for displacement to adjacent sites and for the effect wearing off, which is a sign the system has become predictable.

Do mobile surveillance trailers prevent crime or just record it?

It depends entirely on the trailer. A trailer with cameras alone is a recording device on wheels and inherits the weak deterrent effect of passive CCTV. A trailer designed for prevention is visible from the approach, runs AI that fires on the intrusion rather than on everything, responds to the specific person with light, strobe, siren and a talk-down, and escalates to live operators with a dispatch path. Sharpvue's MAST, MAST-mini, PSU and SPSU units are built around that sequence; the cloud recording, fleet portal and diagnostics are the layer on top.

What does it cost a business to prosecute a theft?

There is no single published figure, which is part of the problem the TalkLP episode raises: the cost of catching is spread across budgets nobody adds up. For the business it includes loss prevention and management hours on the case, employee time in court, downtime while the item is replaced, insurance deductibles and premium increases, and restitution that is ordered but rarely collected in full. For small-value cases the business's own cost of pursuing the case routinely exceeds the loss.

Prevention vs. prosecution

An ounce of prevention is worth a pound of prosecution

Catching the thief feels like winning. Then you add up the detective hours, the court days, the downtime and the deductible, notice that only one property crime in twenty ends in an arrest anyway, and the math says the business won the day the thief turned around at the gate.

In short

01Catching a thief is a bad proxy for protecting a business. The loss is already on the books when the investigation starts, and most of the chain that follows is outside your control.
02Of 100 property crimes, about 31 are reported to police and about 5 are cleared by an arrest. Fewer still end in a conviction with restitution actually paid.
03The justice system spends roughly three times what a burglary victim lost just to process the case, and about six times what a shoplifter took. None of that money comes back to the victim.
04Deterrence only works when the offender sees the measure, understands it is live, and believes a consequence follows. Cameras that only record fail the second and third tests.
05Prevention is measurable: baseline your incident rate, date the deployment, count again. That number, not an arrest count, is the return on a surveillance trailer.

Where this comes from

Two TalkLP episodes, one uncomfortable question

In the autumn of 2026 TalkLP host Amber Bradley ran a two-part series on deterrence with Steve Lindsey, co-founder and chief technologist of the surveillance company LVT. The first, What actually IS Deterrence?, is about why some deterrents keep working and others fade in weeks. The second, An Ounce of Prevention, a Pound of Prosecution, asks the question loss prevention rarely asks out loud: you caught the offender and built the case, but did the business actually come out ahead?

The argument is simple. The industry measures wins by apprehensions and case values because those are easy to count and good for a career. The money a business did not lose because the incident never happened is hard to count and good for nobody's résumé. So the budget flows toward catching, and the cost of catching is spread across so many ledgers that nobody adds it up.

This guide adds it up. It is written by Sharpvue, which builds surveillance trailers whose entire reason for existing is the second path below, so read the final section as a point of view. Everything before it is sourced, and the sources are listed at the end.

Two ways to count a win

The chain that starts after the loss, and the one that ends before it

Every property crime can be told as one of two stories. In the first, the loss happens and a chain of hand-offs begins: somebody discovers it, somebody reports it, a detective picks it up, an arrest may follow, a prosecutor may take it, a court may order restitution. Each hand-off takes time, costs money and loses probability. In the second story, the offender arrives, notices the site is protected, understands the protection is live, believes something will happen to them, and leaves. There is nothing to discover, report, investigate or prosecute.

The first chain is the one the industry knows how to measure. The second is the one that pays.

The after-the-fact chain has seven hand-offs, most of them outside the business. The before-the-fact chain has four, all on the site. Only one of them ends with the loss still on the books.

How often catching works

Of 100 property crimes, about five end in an arrest

Start with the odds. In 2024, US law enforcement cleared 15.9%of reported property crimes by arrest or exceptional means, against 43.8% of violent crimes, according to the FBI's summary of reported crime. By offense the picture is worse where it matters to a jobsite or a lot: about 15% of burglaries, 17% of larceny-thefts and 9% of motor vehicle thefts were cleared.

Those are rates of reported crime. The National Crime Victimization Survey puts the share of property crimes that households report to police at roughly three in ten, and the ones that do get reported are often reported for the insurance file rather than in any expectation of a result. Put the two together and about five of every hundred property crimes end in an arrest. A clearance is not a conviction, a conviction is not a restitution order, and a restitution order is not money.

The catch-it-after funnel, each bar to scale against the first. Reporting rate from the National Crime Victimization Survey (households) and clearance rate from the FBI, both 2024; a clearance is an arrest or exceptional means, not a conviction. The last bar is dashed because no national figure follows a property crime all the way to restitution actually paid.

None of this is a complaint about police. Property crime scenes get little forensic processing, labs rightly prioritise violent cases, and detective time is finite. It is simply the arithmetic a business is betting on when it decides that recording the crime is the same as being protected from it.

What a case costs

The system spends more processing a burglary than the burglar took

The most careful attempt to price a single offense is McCollister, French and Fang's The Cost of Crime to Society(2010), which combined victim losses, criminal-justice system costs and the offender's lost productivity into one figure per crime, in 2008 dollars. The criminal-justice line alone, meaning police, courts and corrections, is the striking one. For a household burglary the victim lost $1,362 and the system spent $4,127 processing it. For a larceny the victim lost $480 and the system spent $2,879. Only motor vehicle theft, where the stolen item is expensive, has a victim loss above the cost of the case.

What the victim lostWhat the justice system spent to process it

Per offense, tangible costs in 2008 dollars. The orange bar is public money spent on the case, none of which returns to the victim. Add fifteen years of inflation and the shape does not change.

That is the public bill. The business pays its own, and it is the one Lindsey and Bradley are really talking about when they ask whether the company came out ahead. It never appears as a line item, because it is paid in hours and delays rather than invoices.

Loss prevention and management hours: Pulling video, writing the report, building the case file, meeting the detective, and the follow-up calls. Every hour is an hour not spent on the next site.
Court days: An employee subpoenaed to testify is paid for the day and produces nothing. Continuances mean the day happens more than once.
Downtime and schedule slip: A stolen generator or a stripped copper run stops a trade for days. Idle crews, missed milestones and liquidated damages dwarf the replacement cost.
Insurance: A deductible per claim, a premium that rises with the claim history, and sometimes a carrier that walks away from the account.
Replacement and lead time: The item is rarely in stock, the replacement is at today's price, and the rental to bridge the gap is on top.
Restitution that never arrives: A court order is not a cheque. An offender with no assets pays nothing, and collecting from one who has some takes years.

Add it up for a stolen skid steer, a stripped copper run or a trailer off the dock and the conclusion is the same every time: by the time there is a case to win, the business has already lost. The case is damage control, and the question is only how much more it will cost.

Why cameras stopped being enough

A camera that only records is a bet on the five percent

For decades the camera itself was the deterrent. Lindsey's phrase for what changed is the TikTok era: a generation that is filmed all day has lost its instinct to behave differently on camera. He cites a Harris Poll finding that about three in four people still improve their behaviour when they know they are on camera, which sounds reassuring until you notice that one in four no longer care, and that the one in four is the group you bought the camera for.

The research agrees with him. The Campbell Collaboration's 40-year review of CCTV evaluations found a modest but significant overall reduction in crime, with the largest and most consistent effects in car parks. But the effect depended on how the cameras were used. Actively monitored systems produced significant reductions; passive, record-only systems did not. Schemes that paired cameras with other measures, lighting above all, did better than cameras alone.

A record-only camera asks the offender to fear the after-the-fact chain. They have done the arithmetic above, even if only by experience, and they know the odds. Recording is evidence, and evidence is valuable. It is not prevention.

What deterrence actually is

See it, get it, fear it

The framework Lindsey leans on comes from Dr. Read Hayes and the Loss Prevention Research Council (LPRC), and it is useful because it is written from the offender's side of the fence. For a measure to deter, the offender has to see it, has to get it, meaning understand what it will do to their plan, and has to fear it, meaning believe the consequence is real. Fail any one and the measure is decoration.

Most security spending fails at step two or three. The dome camera under the eave is seen, but nobody believes anyone is watching. The pre-recorded warning that plays for every passer-by is heard, but it is a motion light: by the third visit it is part of the scenery. Lindsey's point about desensitisation is the sharpest one in the episode. A deterrent that works in month one and is ignored by month three has not failed because it was weak. It failed because it was predictable.

The three conditions, where each one usually breaks on a real site, and what holds it. The right-hand column is what a prevention-first unit is designed around.

The second half of the framework is accountability, and Lindsey is blunt that the two depend on each other. Deterrence without consequence collapses, because the offender learns there is none. That is not an argument for prosecuting everyone. It is an argument for a response the offender experiences in the moment: a voice that describes their jacket, a light that follows them, an operator who is plainly real and has already called it in. The consequence is being seen and addressed by a person, now, which is a different thing from a case file in six months.

Measuring what did not happen

Prevention has a number. It is just not an arrest count.

The objection to prevention has always been that you cannot measure a crime that did not occur. You can; you measure it the way every other intervention is measured. Lindsey's method on the episode is the same one a public-health team would use: establish a baseline crime rate for the site, apply the deterrence, and watch the rate. His benchmark for a well-designed deployment is a reduction in the order of 70%.

Incidents per month, baselineIncidents per month, deterrence in placePeriod average

A worked example with illustrative counts. The method is what matters: the same count, the same way, before and after a dated deployment.

How to run it on your own site

A four-step method for measuring prevention
Step	What to do	What to watch for
1. Baseline	Count incidents for a period you trust: thefts, break-ins, trespass, vandalism, near-misses. Use the same source throughout: incident reports, police reports or the system's own event log.	A baseline built from memory. Write it down before the unit arrives.
2. Date the deployment	Record the day the unit went live and what it was configured to do: zones, schedules, responses, who gets the escalation.	Changing three things at once. If lighting, fencing and a trailer all land the same week, you will not know which worked.
3. Count again	Same count, same method, same length of period. Log the events the system responded to as well as the losses that still occurred.	Counting arrests. An arrest is a loss that already happened.
4. Price the gap	Multiply avoided incidents by what one really costs, including the hidden bill above. Compare to the cost of the unit over the same period.	Displacement to the next site, and an effect that fades, which means the response has become predictable and needs changing.

The result is the only honest return-on-investment figure for a surveillance trailer. Vendors, Sharpvue included, will show you case studies. Your own baseline is better.

Where the math is most lopsided

Jobsites, lots, docks and yards

The prevention case is strongest wherever the stolen item is liquid, the site is empty at night, and the recovery rate is close to zero. That describes most of the places a mobile surveillance unit gets parked.

Construction sites

Theft from construction sites costs the US industry about $1 billion a year in direct losses. In a study of more than 15,000 NIBRS incidents, contractors lost about $6,000 per incident, trucks about $42,000, and under 7% of stolen items were ever recovered. That is not a problem the arrest-and-prosecute chain is going to solve.

Parking lots and dealerships

The one setting where camera research is unambiguous. The Campbell Collaboration's review of CCTV studies found a 51% drop in crime in car parks, and the 2019 update found the largest and most consistent effects there, with actively monitored systems outperforming passive ones.

Docks, yards and cargo

As Lindsey puts it on the episode, a whole trailer can vanish off a dock and nobody knows for hours. By the time it is reported the load is unloaded. The only version of this story with a good ending is the one where the truck never backed up to the dock.

Equipment and material yards

Copper, fuel, catalytic converters and attachments are liquid the same night. Recovery rates for these categories are the lowest of all, and the case value is often below the threshold a prosecutor will take on.

Where Sharpvue stands

We built the unit for the second path

Everything Sharpvue makes exists to stop the incident before it is an incident. The MAST (Mobile AI Surveillance Trailer), the MAST-mini, the pole-mounted PSU and SPSU all run Sharpvue OS, which manages cameras, connectivity, power and deterrence as one system because deterrence is the job, not an accessory. The cloud recording, the fleet portal, the diagnostics and the site planner are genuinely useful, and they are the cherry on top. The cake is the offender who left.

Built to be seen

A mast that rises above the fence line with lighting that reads as live from the road. The unit is parked where the approach is, not where the power is, because it carries its own.

AI that watches the line, not the whole lot

VCA 2.0 analytics at the edge fire on intrusion, line-crossing and loitering in the zones you draw. A person at the gate after hours is an event. A raccoon is not.

A response that happens to the offender

Light, strobe and siren fire on the event, not on a timer. The response is tied to the person in frame, which is the difference between a cue and a motion light.

A voice with a name for them

Talk-down through the mast speaker, automated or from a live operator who can describe the jacket and the truck. A message aimed at one person is not the pre-recorded loop everybody tunes out.

Live operators and a dispatch path

Remote video monitoring puts a person on the event when the AI escalates it, with a documented dispatch path per site. That is the consequence that makes the voice credible.

A system that changes before the offender adapts

Rules, zones, schedules and responses are tunable per site and per hour over the air. The unit that worked in month one is not the same unit in month three.

Build a Trailer Remote video monitoring See the trailer specs

Open by design: every unit streams standard RTSP and supports ONVIF, with your choice of Sharpvue iVISION, Alarm.com for Business or OpenEye on top, and it drops into any third-party VMS. Prevention should not require a platform change.

Frequently asked questions

Is it cheaper to prevent crime than to prosecute it?: In almost every property-crime case, yes. The victim's loss is already spent when the investigation starts, and the public cost of processing the case is usually larger than the loss itself: about $4,127 of criminal-justice cost for a burglary where the victim lost $1,362, and about $2,879 for a larceny worth $480, in 2008 dollars. On top of that the business pays its own hidden bill in staff hours, court days, downtime, insurance and replacement. A measure that stops the incident avoids the whole chain.
What percentage of property crimes are actually solved?: In 2024 US law enforcement cleared 15.9% of reported property crimes by arrest or exceptional means, against 43.8% of violent crimes, according to the FBI. By offense the figures are about 15% for burglary, 17% for larceny-theft and 9% for motor vehicle theft. Because only about three in ten property crimes are reported to police in the first place, roughly five of every hundred property crimes end in an arrest, and a clearance is not a conviction.
Does video surveillance deter crime?: It can, under conditions. The Campbell Collaboration's systematic reviews of CCTV found a modest but significant overall reduction, with the largest and most consistent effects in car parks, a 51% drop in the 2008 review. Actively monitored systems produced significant reductions while passive, record-only systems did not, and schemes that paired cameras with other measures such as lighting did better than cameras alone. A camera that only records evidence deters less every year as people learn that recording rarely leads anywhere.
What is the see it, get it, fear it model of deterrence?: A framework from Dr. Read Hayes and the Loss Prevention Research Council (LPRC) for designing deterrents from the offender's point of view. The offender has to see the measure, get what it will do to their plan, and fear it as a credible consequence. A deterrent that fails any one of the three has no effect. Steve Lindsey uses it on the TalkLP deterrence episode to explain why visible, responsive, consequence-backed systems keep working while passive cameras fade.
How do you measure crime prevention if nothing happened?: The same way you measure any intervention: establish a baseline incident rate for the site over a period you trust, date the deployment, then count the same way afterwards. Count incidents, not arrests, and include near-misses and trespass events the system logged. Lindsey suggests looking for a reduction in the order of 70% from a well-designed deterrence deployment. Watch for displacement to adjacent sites and for the effect wearing off, which is a sign the system has become predictable.
Do mobile surveillance trailers prevent crime or just record it?: It depends entirely on the trailer. A trailer with cameras alone is a recording device on wheels and inherits the weak deterrent effect of passive CCTV. A trailer designed for prevention is visible from the approach, runs AI that fires on the intrusion rather than on everything, responds to the specific person with light, strobe, siren and a talk-down, and escalates to live operators with a dispatch path. Sharpvue's MAST, MAST-mini, PSU and SPSU units are built around that sequence; the cloud recording, fleet portal and diagnostics are the layer on top.
What does it cost a business to prosecute a theft?: There is no single published figure, which is part of the problem the TalkLP episode raises: the cost of catching is spread across budgets nobody adds up. For the business it includes loss prevention and management hours on the case, employee time in court, downtime while the item is replaced, insurance deductibles and premium increases, and restitution that is ordered but rarely collected in full. For small-value cases the business's own cost of pursuing the case routinely exceeds the loss.

Questions and answers on this page are written to be quoted. If you cite them, cite the sources below too.

Sources

Cost-per-offense figures are stated in the study's 2008 dollars and have not been inflated. Clearance rates describe crimes reported to law enforcement; the reporting rate is from household victimisation data and is used here as the best available proxy for commercial sites. The Harris Poll figures are as cited by the guest on the episode. Last reviewed 2026-11-01.