While retail leaders race to keep up in an omnichannel world, there are two adversaries quietly wrecking profits.

Returns and shrink are doing immense damage to retailers’ bottom line. Last year, consumers made $706B in returns, and retailers faced $89B in loss from shrink, much of it preventable. Both are growing because traditional views on loss prevention are riddled with blind spots.

The challenge for retailers: how do they address these behaviors without damaging customer experience or loyalty? Most are stuck using outdated approaches because they’re scared to rock the boat.

We’ve brought together survey data from 1,000+ consumers, internal analytics from Appriss Retail clients, and industry data to show the extent to which loss is affecting profit margin.

Instead of focusing solely on fraud—an outdated angle on loss prevention–the 2026 TRL report looks at how retailers’ approach to behavior patterns and transactional data can act as both deterrents and profit boosters.

Today, returns impact every part of the organization. Operations teams process the volume. Customer experience teams handle the disputes. Merchandising absorbs the margin hit. Loss prevention can’t solve this in isolation.

Returns added up to $706B in 2025 – a figure that’s grown as omnichannel shopping expands and return policies become more generous. $100B of that is preventable fraud and abuse.

For customers, ease of returns is a major purchase factor. Yet many retailers still rely on one-size-fits-all return policies, leading to frustrated customers and higher fraud risk. Create too much friction, these buyers are happy to take their business elsewhere.

Finding the sweet-spot between protecting margins and keeping good customers isn’t always easy. But, the clues are hidden in how you’re thinking (or not thinking) about returns. 

Product quality and competitive prices drive loyalty, while service, convenience and return experiences reinforce it. 

Understanding what drives loyalty matters because blanket return policies ignore these nuances entirely and risk alienating your best customers.

Assessing customer risk requires analyzing return velocity, cross-channel behavior patterns, product categories, purchase history, and temporal trends—impossible to evaluate manually at the point of sale. Without this analysis, retailers are defaulting to blanket policies that treat all customers the same.

But many of these policies, designed to stop fraud and abuse, punish your most profitable relationships.

The smarter approach: Customer-level decisions that account for risk potential, purchase volume and return rate–not just return dollars–protect margins without alienating your best customers. 

But implementing customer-level decisions requires something most retailers don’t have: unified visibility across all channels. That’s where the omnichannel challenge begins.

55% of customers read return policies before making a purchase. This number rises with return frequency (84% for very frequent returners), and those buyers notice when in-store and online experiences differ.

Regardless of return frequency, shoppers prefer in-person return methods  — especially if they bought the item in store.

While a majority of consumers still return items in store, nearly half have utilized at least one method of omnichannel return. 

Convenience or exposure? 

Each channel operates with different detection capabilities, different policy enforcement, and different fraud and abuse flags. A customer blocked for suspicious activity online can return the same item in-store with no warning. 

With this disconnect, criminals and abusers don’t need sophisticated schemes–they simply move between channels to avoid detection.

Retailers controlling omnichannel fraud and abuse are unifying return data across all channels. 

Risk-scoring and behavioral analytics at the customer-level let retailers spot abuse patterns, prevent duplicate losses, and treat customers consistently across channels.

But to get there, they need to understand where fraud actually lives, and why customers are so susceptible to inadvertently abusing the system.

Retail leaders are forced to weigh gaps in their detection capabilities against their appetite for risk. They don’t want to alienate customers with stringent enforcement policies, but the alternative is to “turn a blind eye” to abuse. 

Without visibility into the customer behaviors that constitute fraud and abuse, any picture of Total Retail Loss remains incomplete. 

The data shows that generous return policies can boost customer loyalty—but they also drive higher return rates and abuse when boundaries aren’t clearly communicated. Reducing total returns means changing these behavior patterns, not just catching fraud after the fact. And it has to be done with surgical precision, so you don’t alienate your best customers with unnecessary denials.

Many customers expressed comfort with (sometimes multiple) gray-area behaviors. 

Our data shows a tolerance gap between occasional and frequent returners (one that could and should impact your policy boundaries).

Most consumers engaging in abusive return behavior don’t realize they’re doing anything wrong. They’re operating within what they perceive as acceptable boundaries— whether returning a high percentage of their purchases, buying multiple sizes/colors (known as “bracketing”), or engaging in rental behavior (commonly known as “wardrobing” in apparel). 

This creates a challenge for retail leaders: how do you address behavior that represents a significant portion of loss without damaging the customer experience?

The answer isn’t stricter enforcement across the board. It’s targeted intervention that educates before it punishes so customers can course-correct before relationships (and margins) deteriorate.

Return fraud doesn’t just impact loss. It simultaneously degrades revenue, cash flow, margin, and inventory accuracy. In our data we found fraud concentrated in two distinct zones:

Non-receipted returns: 9.6% of dollar volume ($68B), but home to the most blatant fraud 

Receipted returns: 90.4% of volume ($638B), where you find marginally profitable customers.

Making policy adjustments like “no receipt/no return” may lower non-receipted returns but the collateral damage from denying good consumers far outweighs the benefits. Professional fraudsters will socially engineer the associate or call center to give them the return anyway via an override.

The goal isn’t just to stop bad actors, it’s to protect honest shoppers, deter abuse, and help associates make better, more consistent decisions. 

Instead of: responding to margin pressure by reintroducing return fees across the board.

Try: maintaining generous policies for the majority and using AI decisioning to apply graduated controls only to high-risk behaviors.

High-return customers are more likely to push policy boundaries. The majority (62%) won’t commit outright fraud, but in the meantime, your generous return policy becomes a subsidy for serial abusers.

The problem takes shape when customers’ return rates grow faster than their purchase rates. Regardless of spend, this behavior shouldn’t be encouraged, as the rate of return makes them unprofitable.

Warn and approve creates a three-tier approach to return decisions. Unlike traditional systems that only approve or deny returns, this adds a critical middle tier: the warning.

Approve: Legitimate returns from low-risk customers proceed without friction

Warn: Customers showing high-risk patterns receive a notification before facing a decline, giving them the opportunity to course-correct

Decline: Returns that meet fraud or extreme abuse thresholds are blocked

This graduated response educates customers before enforcing consequences in order to preserve relationships while reducing abuse.

Mind the age gap

Most shoppers (61%) are open to tech-driven return eligibility checks. The transparency customers demand isn’t fear-based—they’ve accepted AI as part of transactions, they just want to understand how decisions are made.

Warn and approve protects associates from the burden of making difficult return decisions. Here’s why it works:

The system identifies risk. The associate no longer has to be “the face” of return decisions. They can direct customers to a phone number or website for appeals, helping them avoid in-person friction when disputes arise.

Humans are error-prone and vulnerable to bias. Without visibility into the entire customer profile, it’s impossible to make the right call every time. When you remove the burden of judgment from the associate–you remove the conflict. 

Clear explanations and escalation paths let retailers capture AI efficiency without eroding trust. 

Customers who leave their returns experience with a smile cite the same qualities: fast, friendly and free. 

Creating this kind of positive experience at scale requires the ability to make consistent, risk-based decisions across millions of transactions.

Massive volumes of transactional data make spotting shrink, fraud, operational errors, and inefficiencies slow and resource-intensive, allowing losses to compound while teams investigate. 

When data doesn’t flow between channels, retailers face a double problem: they can’t identify root causes of shrink and margin leakage, and they can’t see what’s happening in real time. Both organized criminals and opportunistic employees are able to exploit those gaps.

Total shrink sits at $90B across all channels, but the operational damage extends well beyond missing inventory. Employee theft, operational errors and ORC add up to $47B in total shrink for retailers. 

$26B | 29% of shrink

Associates processing fraudulent transactions create phantom inventory, so where the system expects ten units, only two remain. Fake refunds, unauthorized voids, and collusion schemes drain both sales and margins while distorting the data retailers use for ordering decisions.

$19B | 21% of shrink

Distribution center mispicks and transfer errors create a cascade of inaccurate counts. These drive out-of-stock conditions that cost sales and erode brand loyalty, while losses accumulate undetected. Mismatched pairs or sets rendered unsellable add to the damage.

$12B | 13% of shrink

Perishable spoilage. Damaged merchandise. Missed recalls. Coupon abuse. Price manipulation. In some categories, operational losses account for up to 70% of store shrink–waste that goes unnoticed until it shows up on the P&L.

Shoplifting, particularly in organized retail crime, surged by double digits in the past year and now accounts for $9B in total shrink. Internal theft has also increased. These issues create safety concerns that drive customers away and out-of-stock conditions that cost sales, impacting both revenue and brand loyalty. 

Over a quarter of shoppers (27.6%) would stop buying from a retailer if they felt unsafe in stores.

Addressing shrink means connecting three systems that typically operate independently:

Traditional forensic systems are one-dimensional. They only identify things like employee theft, and often only one behavior at a time. Meanwhile, retailers might face thousands of dollars in losses over months and years of “investigation.” With AI-enhanced analytics, retailers get insight into issues as they occur.

These AI systems connect POS data, inventory counts, and transaction patterns to determine whether high shrink is internal fraud versus external theft, eliminating the guesswork about where to focus resources.

As this report has shown, loss prevention now extends far beyond traditional ideas of “shrink.” Returns fraud and abuse are the bigger issue–as customers often don’t even realize they’re committing these abusive behaviors (but won’t go running if you warn them that they are). 

Retailers are already protecting profits without sacrificing customer experience by taking a Total Retail Loss approach. 

Loss prevention is no longer a department-level effort. It’s an enterprise strategy — bringing together loss prevention, operations, finance, and customer experience under a shared mandate: reduce total loss while protecting margin and loyalty.

This next wave of loss prevention will merge two capabilities most retailers still treat separately:

Retailers who unify these will spot abuse earlier, protect high-value relationships, and maintain friction-free experiences that build long-term loyalty. Fraud thrives in the gaps between systems. In an omnichannel world, unified data is the best defense. 

Modern TRL solutions combine two types of AI:

Real-time decisioning based on customer risk models stops fraud as it happens—approving legitimate returns, warning high-risk behaviors, and declining abuse at the point of sale.

Forensic AI within loss analytics identifies root causes after the fact—surfacing patterns across employees, locations, products, and processes to prevent future loss.

Together, these capabilities let retailers significantly reduce:

• High processing costs
• Time spent decisioning on returns
• Complexity of returns analytics
• Bracketing/wardrobing, fraud, and inconsistent shopper experiences

For returns specifically, retailers implementing AI need to take an approach that allows them to progressively optimize AI-decision-making thresholds without harming customer relationships. 

Day One: Start with data gathering – act on clear cases while placing uncertain cases on “forced approval” to collect behavioral data

Early Learnings: Use AI to identify patterns and quantify impact before making any changes to operational processes and procedures. 

Collaborative Evolution: Make data-based decisions to calibrate thresholds gradually.

This process isn’t about dramatic overnight changes but about evidence-based, incremental improvements that build retailer confidence over time.

If you’re ready to move beyond reactive loss prevention and build a TRL strategy that protects both margins and relationships, let’s talk.

Download the full report and share.

Run the numbers for your category with our Total Retail Loss Calculator.

Customer survey

We conducted an online quantitative study of 1,020 U.S. customers aged 18+ who made at least one return in the past year. The survey ran in [December, 2025]. Respondents were balanced across gender, generation, and return frequency to ensure representative coverage of occasional, frequent, and very frequent returners.

Survey responses were weighted to reflect U.S. population demographics and validated against known industry patterns. Where customer self-reporting diverged from observed behavior in Appriss systems, we noted the gap and prioritized observed data for financial projections.

Appriss customer data

Actual transactions and internal analytics from 250 million unique customer identifiers across Appriss Retail clients provided real-world validation of customer-reported behaviors. This data informed our estimates on fraud rates, abuse patterns, processing costs, and the effectiveness of warn and approve interventions.

Industry benchmarks

We referenced available data points from 2022 through 2025 on shrink rates, organized retail crime, and omnichannel adoption from The National Retail Federation (NRF), Appriss Retail internal stats, as well as inventory returns research and publicly available data from ECR Retail Loss, IHL Group, and Deloitte. We used this data to contextualize our findings within the broader TRL environment.