Jewelry
·
Lookback
Strong retention was hiding $2.3M in return abuse and $2.4M of loyalty-driven growth.
A 12-month lookback across 850K customers for an omnichannel jewelry brand. The best customers were loyal. A small group of the rest was costing millions.
VIP growth upside
CX review hours a year to automate
model accuracy
The challenge
2,600 hours a year spent inspecting returns by hand
This omnichannel jewelry brand has the kind of retention most retailers want, with about 850K customers and a core of loyalists who keep coming back. Jewelry is also an easy category to abuse: items are small, valuable and hard to verify on return. About 14% of sales by value came back, and inside that flow were fake and empty returns, wardrobing, item swaps and a group of repeat bad actors. The CX team had no automated way to separate genuine returns from abusive ones, so they inspected them manually, at a cost of 2,600 hours a year. At roughly 30% gross margin, every dollar lost to abuse needed more than three dollars of new sales to replace.
What Pinch found
$2.3M in return abuse sat beneath strong retention
Most of the loss came from a middle group rather than outright fraud. Suspect abusers, customers with extreme return rates and behavior that did not match their purchase history, accounted for the bulk of it. The egregious cases, empty boxes and wardrobed pieces, were smaller but clear-cut. Promo abuse ran through accounts with subtle email variations reusing one-time codes across shared addresses, which stays invisible unless accounts are linked. On the other side sat loyalists buying about $30M a year with no differentiated experience. At 30% margin, offsetting the abuse alone would take about $8M in new sales.
Suspect abusers with extreme return rates and inconsistent behavior.
Egregious abuse: fake or empty returns and wardrobing.
Promo abuse. Accounts with subtle email variations reusing one-time codes across shared addresses.
VIP upside from exchange rebates and premium handling for loyalists who buy about $30M a year.
Most of the loss sat with suspect abusers, not the obvious fraud. The upside from loyalists was about as large as all the abuse combined.
The response
Stop repeat abusers, give loyalists a reason to stay
Pinch recommended matching the response to the evidence. Egregious abusers would be denied at return initiation and lose future orders. Suspect returners would see restocking fees, photo checks and in-store inspection, which slows abuse without accusing anyone outright. Linked accounts would be stopped from reusing one-time codes. Loyalists would get instant refunds at label scan, exchange rebates and premium support, worth an estimated $2.4M in growth. The 2,600 hours of manual review would be automated.
Egregious abusers
Block
Deny returns at initiation and cancel future orders.
Suspect returners
Friction
Restocking fees, photo checks and in-store inspection.
Promo abuse
Block
Stop linked accounts from reusing codes.
Best customers
Reward
Instant refunds at label scan, exchange rebates and premium support.
How Pinch worked
100 behavioral features scored on every return
Pinch ID clustered emails, addresses, phones, devices and payment methods so that one customer with several accounts was treated as one. Ensemble models then scored each return on 100 behavioral features, classifying returns at 98% accuracy and catching 9 in 10 abusive ones. A separate promo layer grouped the linked accounts behind one-time code reuse. Decisions were designed to flow into the brand's order, returns and warehouse workflows, so a flag turns into an action rather than a report.
Pinch ID
Identity clustering across emails, addresses, phones, devices and payment methods.
Ensemble abuse models
100 behavioral features scored per return.
VIP uplift
Loyalists get exchange rebates and premium handling.
Promo detection
Linked accounts clustered and decisions pushed into order, returns and warehouse workflows.
The takeaway
Brand details are anonymized. Figures come from the brand’s own order and return data.
Industry
Jewelry
Size
$100M+ online sales
Platform
Shopify
Solutions
Return Abuse
Lookback
12 months
At a glance
~14%
Return rate by value
~30%
Gross margin
850K+
Customers
Online orders
Scope of the lookback
More customer stories
Apparel
·
Lookback
projected ROI
A 10% return rate hid three abuse patterns. Fixing them projects to $230K+ a year at 5x ROI.
Read the story →
Home
·
Lookback
exposure found in 24 months
A 4% return rate on $1B+ of sales was hiding $38M+ in reseller and abuse exposure.
Read the story →
Automotive
·
Lookback
abuse identified in 12 months
A 6% return rate looked unremarkable until identity clustering exposed $1.5M+ in claims fraud and serial returns.
Read the story →
Free lookback · for qualified brands
See what you lost last year. Free.
01
Share 12 months of data.
Install the Shopify app or SFCC cartridge, or send order, return and claim exports.
02
Get your lookback in under two weeks.
What you lost to return and claims abuse, and who took it.
03
Qualified brands try Pinch free for three months.
Run it on live orders, then decide.