Most DTC brands assume they’ll see fraud when it happens.
But what if the biggest losses don’t look like fraud at all?
This case study breaks down how one lean, high-margin footwear brand uncovered a hidden return fraud problem - one that wasn’t obvious in their reports, wasn’t flagged by their support team, and wasn’t caught by their tools.
They didn’t rebuild their tech stack or hire more people. Instead, with the help of Jordan at YoFi, they added one layer: risk scoring. And used it to quietly reshape how decisions got made across returns, CX, support, and marketing.
If you’ve ever had that gut feeling that something’s off in your returns process, but can’t quite prove it, this story is for you.
Let’s dive in. 👇
What’s Inside
The Setup → The fraud they didn’t see coming
The Diagnostic → A 20-min scan that changed everything
The Fix → Risk scores, refund rules, and support routing
The Results → 90% less fraud, 70% faster support
The Playbook → How to replicate this in your ops
Full Q&A → With Yofi’s co-founder, Jordan Shamir
The Fraud They Didn’t See Coming
The footwear brand wasn’t worried about fraud.
They had all the typical guardrails in place: no free shipping, minimal discounting, and a tight focus on margins. Unlike most DTC brands chasing volume at any cost, their whole strategy was about protecting the bottom line.
So when Yofi offered them a free “bad actor diagnostic,” the team didn’t jump. They assumed fraud wasn’t a major issue. A few edge cases here and there - sure. But nothing worth spinning up a new tool.
They were wrong.
What looked like healthy returns data was actually hiding a serious leak. Customers were abusing return windows, cycling discount codes, and even using fake tracking labels to get refunded without sending anything back.
And because it was all happening in the background, across support, shipping, and multiple brands, they couldn’t see the full picture. Until the numbers showed up in black and white.
Proving The Problem Exists
Jordan Shamir, Yofi’s co-founder, knew from experience that return fraud often flies under the radar. Most brands only catch it when it gets so bad that finance starts asking questions.
So instead of pitching features, he asked for data.
Within 20 minutes, the Yofi team connected to the footwear brand’s Shopify store, Loop Returns account, and support tools. Then they ran a scan on historical return and support activity.
The result was a list no one expected.
Some customers were returning multiple orders per month - often after obvious wear.
Others were using fake or rerouted shipping labels to trigger refunds.
A few high spenders looked like VIPs until the return behavior told a different story. One customer had been flagged as “loyal” for years. But once returns, discounts, and shipping costs were added up, they were deeply unprofitable.
“A lot of brands think their best customers are the ones who spend the most. But if they’re returning half their orders, they might be your biggest liability.”
Acting On The Signal
Once the footwear brand saw the scope of the problem, they didn’t debate it. They moved.
Implementation took less than 30 minutes.
They installed Yofi’s Shopify app, plugged in their Loop API key, and immediately started generating real-time risk scores for every customer and return.
These scores weren’t generic.
Each one pulled in dozens of signals - from order and return history to support interactions and even the specific language used in refund requests.
But the power wasn’t just in the score. It was in what they did with it:
1. Pausing refunds until verification
Before Yofi, the brand’s return system was built on speed.
Refunds were issued the moment a return was marked “delivered” by the carrier - no questions asked.
That worked fine for honest customers. But it created a giant loophole for fraud.
Some people were sending back empty boxes. Others were manipulating labels so a package looked like it was headed to the right zip code - but actually went to a neighbor’s house or empty lot.
As long as the tracking said “delivered,” the refund went through.
Yofi helped them change that.
Now, refunds are delayed until verification: either a warehouse scan, a condition check, or cross-referenced data from the carrier. That small change, adding friction only where it’s needed, prevented tens of thousands in losses during the holiday season alone.
2. Creating custom risk profiles
The scores also helped the team identify different types of risky behavior.
Yofi worked with the footwear brand to create separate profiles for fraud patterns they were seeing in the data. For example:
“Bad resellers” were customers constantly using promo codes to cycle through new purchases
“Cyclical returners” wore products during peak season, then sent them back
“Ghost clusters” were networks of accounts tied to the same person - used to bypass bans or order limits
This went beyond blocking one email address. Yofi’s clustering system linked behavior across identities. So if someone got banned on one account, the others tied to that behavior were flagged too.
“You used to be able to block [email protected]. Now we block the entire cluster.”
3. Routing support based on risk
Even the best return rules can be undone in one place: customer support.
Support agents are measured on speed and satisfaction. That means they’re incentivized to resolve complaints quickly - and fraudsters know it.
One emotional ticket can override any policy.
To fix this, the team embedded Yofi’s risk scores directly into their support flow.
Now, support tickets from low-risk customers go to an AI agent. High-risk ones go to a human. The goal isn’t to slow things down - it’s to make sure that exceptions don’t get abused.
“If it’s Yofi high risk, you talk to a human. If it’s low risk, you talk to an AI agent.”
That simple routing change cut fraud-related investigation time by 70%.
4. Personalizing operations based on trust
Once the scores were flowing through Shopify as customer tags, the team started using them to shape customer experience more broadly.
They didn’t just block fraud. They built smarter, more contextual policies:
High-risk customers were charged return fees and had refunds delayed
Low-risk customers got waived fees and faster support
Risky profiles were excluded from email and SMS campaigns entirely
This wasn’t just about protecting margins. It was about making better use of every customer interaction.
Instead of treating everyone the same, they calibrated trust - adding friction where it saved money, and removing it where it built loyalty.
5. From fraud detection to operational signal
What started as a fraud prevention tool quickly became something more useful: a signal that flowed across the stack.
Returns. Support. CX. Marketing.
Every team had access to the same layer of intelligence, and they could act on it without needing a new tool or workflow. The business got smarter - without getting heavier.
And that’s where the real leverage came from.
What Changed
After just a few months, the impact was clear:
90% drop in return fraud
3-4% of total returns declined for fraud
70% less time spent investigating fraud claims
Hundreds of thousands in annual savings
20 minutes to implement, 2 hours to train the team
No new headcount. No complex systems overhaul. Just better data and a faster way to act on it.
The Bottom Line
Fraud isn’t just a payments problem. It’s increasingly a customer support problem - and most teams don’t have the visibility or tools to fight it.
Most brands still treat returns and refunds like background noise. But buried inside those workflows are behaviors that quietly erode your margins. If you’re not tracking them, you’re paying for them.
“Even if you don’t think fraud is happening. It is happening. And this is one of the fastest ways to improve your margins without changing your tech stack.”
The signal is already there. You just need the right tool to surface it - and the willingness to act.
How You Can Replicate Their Success
If you’re managing returns, CX, or customer support, and you’ve got a gut sense that something’s off, this playbook is for you.
This footwear brand didn’t think they had a fraud problem. They weren’t giving away free shipping. Their return policies were tight. But a quick diagnostic showed a hidden leak in the system - one that was quietly costing them hundreds of thousands a year.
They didn’t rebuild their tech stack. They didn’t hire a team of fraud analysts. Instead, they added a risk scoring layer that slotted into their existing tools and gave every decision - from refunds to support tickets to promotions - more context.
Here’s a step-by-step breakdown of what they did, and how you can apply the same playbook to your brand.
1. Quantify what fraud is actually costing you
The team didn’t go looking for fraud. They just wanted to validate that their systems were working. It turned out they weren’t.
Once they pulled the data, it became obvious: refunds were being issued for items that never arrived, some customers had extremely high return rates, and in many cases, delivery scans didn’t match warehouse intake.
What to do:
Pull 6–12 months of return data, customer service tickets, and refund timestamps
Look for patterns: high refund volume without warehouse confirmation, repeat customers with aggressive return behavior, tracking scans that don’t match actual product intake
Use a fraud detection partner or internal analyst to identify and cluster risky behaviors
Why it works:
You can’t fix margin leaks you’re not measuring. This step gives you a clean baseline and helps justify action.
When to do it:
If your return rate is rising, or if you're refunding before returns are physically received.
2. Tie every return and support ticket to a risk score
Instead of overhauling their tools, the brand added intelligence to what they already had. A customer’s risk score - based on order behavior, return activity, and support interactions - flowed into every touchpoint.
What to do:
Use software like Yofi that integrates with Shopify, Loop, and your helpdesk
Score customers based on things like repeat returns, label manipulation, coupon abuse, and suspicious claim language
Pipe those scores into your existing tools using tags or metadata
Why it works:
It gives your team a trust signal for every customer interaction - so they can automate better decisions across returns, support, and fulfillment.
Heads up:
Scoring is only as good as your data. Messy inputs mean unreliable signals.
2A. Align teams around what fraud looks like
Once you start scoring customers, everyone in your org needs to know what that means - and what to do about it.
What to do:
Run a short internal walkthrough for CX, support, operations, and finance
Share examples of “high-risk” vs. “low-risk” customer behavior
Document policies for how scores should affect decisions (refunds, fees, escalation, etc.)
Why it works:
Scoring only creates leverage when teams are aligned. Otherwise, it just becomes background noise.
When to do it:
Immediately after scores go live - so you avoid confusion and missteps.
3. Stop refunding just because a return says “delivered”
This was the brand’s biggest unlock. Previously, refunds were triggered the moment a return showed up as “delivered” in the carrier system - even if nothing actually reached the warehouse.
Fraudsters exploited this by rerouting packages, sending empty boxes, or manipulating labels.
What to do:
Change refund triggers from “carrier delivery” to “item verified at warehouse”
Flag any returns marked “delivered” that haven’t been scanned in
Use audit tools to detect return labels that redirect to incorrect or nearby addresses
Why it works:
Return fraud often hinges on timing. Delaying refunds until verification removes the incentive and breaks the tactic.
When to do it:
Immediately. This one change saved the brand tens of thousands during peak season alone.
4. Route support tickets based on customer risk
Support teams are often the weakest link. Most agents are incentivized to close tickets fast - so fraudsters know they can often get around your rules just by asking nicely.
What to do:
Route low-risk customers to AI or automated workflows
Route high-risk tickets to trained agents for manual review
Set rules that prevent frontline agents from issuing refunds without escalation for risky profiles
Why it works:
Support is where most refund abuse slips through. Adding the right friction protects margins - without hurting good customers.
Common mistake:
Letting agents override policies without knowing whether the customer is trustworthy.
5. Adjust return policies based on customer behavior
Not all customers should be treated the same. This brand gave instant refunds and waived fees for trusted customers and made serial returners wait or pay.
What to do:
Use Shopify tags (or your OMS) to segment customers by risk
Set different refund speeds, return windows, or fees based on the customer’s trust level
Keep logic internal - don’t announce thresholds or policies publicly
Why it works:
This aligns your return policy with your profit model. Loyal customers get better service. Problematic ones stop gaming the system.
When to do it:
As soon as scoring is in place. The value isn’t in knowing the score - it’s in using it to drive action.
5A. Use fraud insights to refine your incentives
Once you’ve segmented your return policy, take it a step further. Use the same insights to rethink how you design promotions, perks, and loyalty.
What to do:
Reassess loyalty tiers - are high-return customers getting early access or exclusive offers?
Exclude high-risk profiles from discount campaigns or upsell flows
Surprise low-risk customers with faster fulfillment or waived fees
Why it works:
Risk data isn’t just for defense. It helps you stop rewarding the wrong behaviors - and start doubling down on the right ones.
When to do it:
Once your policies have stabilized and scores are flowing into your marketing and CX systems.
6. Share fraud signals across teams and brands
The footwear brand is part of a portfolio that includes other consumer brands. Instead of operating in silos, they linked risk signals across stores.
What to do:
Combine customer data across brands or banners
Track identity clusters (not just email addresses)
Flag and suppress known bad actors in support, marketing, and fulfillment
Why it works:
Fraudsters often hit multiple stores. Sharing data across teams ensures they don’t get a second chance.
Common mistake:
Thinking a new email address = a new customer.
7. Start small: fix the most painful leak first
The brand didn’t try to fix everything. They picked one margin-killing behavior - refund fraud - and focused entirely on that. Once they saw the ROI, it became easy to scale.
What to do:
Find the single highest-cost failure in your return flow
Add risk scoring and routing to stop just that issue
Prove the ROI before expanding to other areas
Why it works:
Trying to fix everything slows you down. Fixing one thing well gives you momentum, proof, and internal buy-in.
When to do it:
Now. Most operators wait too long trying to build a full strategy. Just fix the part that’s bleeding cash today.
8. Build a feedback loop to keep your fraud model sharp
Fraud tactics evolve. So should your system.
What to do:
Run a quarterly fraud review with your CX, Ops, Finance, and Support teams
Look at: top fraud patterns blocked, score drift, false positives, support trends
Use those insights to adjust thresholds, workflows, or escalation rules
Why it works:
Fraud prevention is not a set-it-and-forget-it system. Feedback loops keep it effective and help you discover new places to apply the signal.
When to do it:
Every 3-4 months, or after major launches, sales events, or policy changes.
Q&A With Jordan Shamir
If you want to go deeper on this project, here’s the full conversation I had with Jordan Shamir, Co-Founder of Yofi.
We walk through how a DTC footwear brand uncovered hidden fraud inside their returns process, and the exact steps they took to stop it: from real-time risk scoring and refund controls to support routing and customer segmentation. Jordan explains how they integrated everything in under 20 minutes, used risk signals to block bad actors, and helped the brand save hundreds of thousands without changing their tech stack or adding headcount.
If you’re responsible for returns, customer experience, or operational margin, and want to tighten up leak points without slowing down your best customers, this is a tactical breakdown worth studying.
Table of Contents:
This conversation has been edited for length and clarity.
1. The Setup
How did the conversation with the brand get started?
We met them at a conference. At first, they were a little skeptical. They didn’t think they had a fraud problem. This brand is very profitability-driven. They don’t do free shipping. They rarely discount. Their whole strategy is about protecting margins - so their mindset was like, “We’ve already got a tight ship.”
They were doing a few things to fight fraud already, but it was manual. They were literally looking people up on Whitepages. So we offered to run our bad actor report - that’s our internal diagnostic where we pull in return data and quantify fraud across their order history.
That alone was eye-opening. They realized a lot of returns weren’t even making it back to the warehouse. People were taking advantage of free gifts and promotions. And when they saw that it wasn’t just one-off issues - it was happening consistently - that’s when they decided to dig deeper.
2. The Assessment
What made the issue feel urgent for them?
A couple things. First, we showed them very specific examples. These weren’t hypotheticals - these were real customers sending empty boxes or manipulating tracking labels. Once they saw that, it was hard to ignore.
Second, we helped them calculate what it was actually costing. Most teams think of fraud as just chargebacks, but post-purchase fraud - claims, returns, shipping label abuse - adds up fast. When someone wardrobes a pair of flip-flops and returns them after summer, you’re not just losing the product.
You’re losing:
The original shipping cost
Return shipping
Warehouse processing time
Inventory resale value (if any)
And possibly, customer support time too
All of that eats into your bottom line. Most brands have 10–20% margins, if that. So it doesn't take many bad actors to wipe out the profit from hundreds of good customers.
It sounds like this was also about shifting how they viewed customer value.
Exactly. A lot of retailers still define their best customers by how much they spend. But if someone is returning half their orders or constantly hitting up support for credits, they might be your worst customer from a profitability standpoint.
What we’re doing is helping brands shift that thinking - from revenue to contribution margin per customer. It’s a hard shift, but it’s where the industry is heading.
3. The Fix
Once they agreed to move forward, what did the implementation look like?
They already had a strong tech stack - Shopify, Loop Returns, Gorgias, Sierra AI - so we were able to move really quickly.
Here’s what implementation looked like:
They installed our Shopify app
Connected their Loop API key
We started running in heartbeat mode (passive mode in the background)
Within 20 minutes, we were generating real-time risk scores
Then we built specific fraud models based on their customer behavior. The three biggest ones were:
Bad Resellers - people who repeatedly used discounts and coupons
Cyclical returners - customers who kept returning product until they hit an unprofitability threshold
Return fraud - fake tracking numbers, rerouted addresses, empty boxes, etc.
How did you help them stop refund-on-delivery fraud?
That was probably the biggest win. Before, the brand issued refunds as soon as a return showed up as “delivered.” But people were manipulating labels - sending empty boxes or using nearby addresses with the same ZIP code to trick the system.
We helped them delay the refund trigger until the item was actually verified:
Either by scan at the warehouse
Or by confirming package integrity with the carrier
Or by validating the data through fraud rules
That one change - just moving the refund trigger back by a day or two - saved them tens of thousands of dollars, especially during peak season.
4. Embedding the Scores into Daily Ops
What did training and rollout look like for their team?
The technical setup was fast - about 20 minutes. But the real value came in a two-hour onboarding with their team. We walked them through:
What the risk scores meant
How to interpret them
How to embed them into workflows across returns, support, and marketing
After that, they took the scores and did some pretty cool things.
What kind of changes did they make?
They used the scores in multiple ways:
Support routing:
High-risk customers were routed to human agents
Low-risk customers went to AI agents
Policy control:
High-risk customers got return fees and delayed refunds
Low-risk customers had fees waived and got refunds faster
Marketing suppression:
They excluded high-risk customers from Klaviyo campaigns
Used tags in Shopify to segment behavior and block promotions
It became an operating layer. Risk scores weren’t just about fraud anymore - they shaped how support, CX, and marketing worked.
How did they handle identity-level abuse - like customers using multiple emails?
Great question. That’s where we cluster identities. Before, if someone used a new Gmail address, the system couldn’t catch it. But we built models that link similar addresses, payment methods, and behaviors.
So instead of just blocking “[email protected],” you block the entire cluster of known related accounts.
5. The Results
What kind of ROI did the brand see?
Big picture, they saw:
90% reduction in return fraud
3–4% of returns blocked before refund
70% reduction in fraud-related support time
Hundreds of thousands in annual savings - possibly low seven figures
This all happened without adding headcount or rebuilding their stack. They used the tools they already had - just with better signals.
How do you price your product in a case like this?
We’re a standard SaaS subscription based on return volume and order count. Our typical ROI is 7–10x. For this brand, they’re definitely in that range - maybe even better. The value is very measurable.
6. Lessons and Advice
What should other brands take away from this project?
The biggest thing? Don’t assume fraud isn’t happening. It probably is - just in places you’re not looking.
Start with a diagnostic. It’s free. Even if you don’t work with us, you’ll have data. And if you do want to act on it, most brands can get up and running the same day.
If you’re trying to improve margin by 2–5% fast, this is one of the easiest levers. You don’t need to renegotiate logistics contracts or build new infrastructure. Just stop refunding fraudsters. It’s margin you already earned - you’re just giving it away.
Is this setup repeatable for most DTC brands?
Definitely. If you're on Shopify and using Loop or Gorgias, we integrate right in. Implementation is simple. Teams are usually self-sufficient after onboarding. We’re always around to help - but the whole point is to make this low-lift and high-impact.
Fraud scoring is becoming the universal filter. Who gets an email. Who gets free shipping. Who talks to a rep. Who gets a refund. Once that layer is live, everything else gets more efficient.
