I went looking for a workflow gap. I found a $2.3M problem.
While researching how warranty SKUs and costs could move into Supplier Oasis, I mapped the existing legacy warranty process and uncovered a much larger operational gap. Overstock had been issuing refunds and replacements on warranty-covered orders outside the standard 30-day return window without a reliable reimbursement and recovery workflow.
- Role
- Product Manager
- Focus
- Warranty platform, partner integration, ecommerce customer experience, operations
- Scope
- Product warranty and shipping protection
- Users & stakeholders
- Customers, Customer Care, Operations, Merchandising, Finance, Engineering, UX, Analytics, Extend
In 30 seconds
- The problem
- Warranty claims ran on paperwork and weekly spreadsheets. Overstock could refund or replace a warranty-covered order outside the standard 30-day return window with no reliable reimbursement or recovery workflow behind it.
- My vantage point
- Nobody assigned me a warranty project. Operations kept raising warranty cost and process friction during Manage Cost discovery, so I audited the workflow — and the audit became the work.
- What I owned
- The audit and escalation to leadership, the Extend integration, the offer and eligibility model, cross-functional delivery across storefront and post-purchase surfaces, and the auto-opt-in A/B test.
- Outcome
- ~$2.3M in historical first-cost exposure identified across roughly seven years (identified, not recovered), 100% shipping-protection coverage on physical items, ~87% SKU eligibility, and a +30% warranty attachment rate in living room and bedroom furniture.
Identified in discovery
~$2.3M
First-cost exposure associated with refunds and replacements on warranty-covered orders outside the 30-day return window, traced across seven years.
Exposure identified — not a recovery claim
- 100%
- Shipping protection coverage across physical, tangible products
- ~87%
- SKU eligibility for product warranty
- +30%
- Warranty attachment rate increase in living room and bedroom furnitureIncrease in rate, not percentage points
The starting point
01
The project started somewhere else.
During Manage Cost discovery, Operations kept raising friction around warranty management. One roadmap objective was to let warranty SKUs and costs be created and managed through Supplier Oasis.
I started auditing the warranty workflow to understand what would need to move into Manage Product and Manage Cost. Following that workflow end to end exposed a deeper issue than the screen I had started with.
The assignment changed
The discovery elevated warranties into a company priority. I was appointed PM over the warranty experience.
The legacy warranty process
02
Two manual paths, with no direct system integration between them.
Customer side
- 01Eligible product page
- 02Customer adds warranty
- 03Order placed
- 04Pamphlet may arrive separately
- 05Customer keeps receipt
- 06Customer contacts legacy warranty provider
Internal side
- 01Legacy warranty provider provides a template
- 02Operations creates and attaches warranty and category data in legacy tools
- 03Operations pulls weekly warranty-order reports
- 04A spreadsheet is built with order, warranty, and category details
- 05The file is manually sent to the previous warranty partner for tracking, audit, and payment
Very manual. Very spreadsheet.
Following the data
03
The claim could happen. The recovery process could disappear.
7 years
The issue accumulated quietly because claims were relatively infrequent and often occurred long after purchase.
Claims did not always flow through the incumbent warranty provider. A customer who had lost the paperwork or receipt — or simply contacted Overstock first — could reach Customer Care instead.
Even outside Overstock's normal 30-day return window, Customer Care could issue a replacement or refund for an order with warranty coverage. Because the internal process was disconnected, those claims were not consistently routed or recovered through the warranty provider after Overstock resolved them.
I escalated the analysis through leadership. Warranties quickly became a high-priority company initiative, and Legal became involved in reviewing and pursuing the historical issue. The outcome of that recovery effort is not represented here.
A wider problem
04
The problem was bigger than money.
Customer experience
A fragmented claim path depended on physical paperwork and proof of purchase.
Operations
SKU and category setup, weekly reports, spreadsheets, and handoffs were manual.
Data & reporting
Claims handled through the incumbent warranty provider were difficult to see and reconcile across the full lifecycle.
Financial control
Overstock could fulfill a covered claim without a connected reimbursement or recovery process.
Redesigning the model
05
From weekly files to a connected lifecycle with Extend.
Extend supported SDK, webhook, and API-style integration patterns with better reporting. We used placeholder SKUs for warranty tiers and categories while Extend dynamically returned the appropriate offer based on product data.
Conceptual data loop
Overstock sends product and order data
Extend determines and records the appropriate protection
Order, fulfillment, and delivery details move through the integration
Coverage is associated with the customer and order
Claim activity becomes visible across the connected process
Purchase and order details were transmitted for the customer, removing the need to separately register or activate the warranty. Extend described its customer-resolution promise as around 10 minutes; that was Extend's service promise, not an Overstock performance metric.
Customer journey placements
- 01
PDP
- 02
Cart
- 03
Checkout
- 04
Post-purchase
Protection could be surfaced again if a customer skipped an earlier offer. A post-purchase reminder was also developed or planned to trigger before the 30-day post-delivery purchase window closed. Conceptual journey — not a product screenshot.
The offer
06
Two protection products, two different jobs.
Shipping protection
100%
Physical product coverage
- Protects the order from shipping through receipt for covered damage or theft scenarios.
- Eligible across tangible Overstock products.
- Digital products, such as online gift cards, were excluded.
Product warranty
~87%
SKU eligibility
- Covers eligible issues after Overstock's 30-day return period under the warranty terms.
- Extend takes on covered replacement or refund requests after that point.
- Lower-volume categories including decorations, decorative accessories, and window treatments were temporarily deprioritized for experience optimization, A/B tests, and bug fixes.
Cross-functional delivery
07
This touched almost everything.
Surfaces
PDP, cart, checkout
Offer placement wherever a customer could still add protection.
Orders & fulfillment
Order and coverage details moving downstream.
Post-purchase & email
A second chance to add coverage before the window closed.
Teams
UX
Placement and the “Powered by Extend” branding.
Customer Care
Extensive training materials for the new claim path.
Merchandising
Advocated for category coverage and eligibility.
Engineering & Analytics
Connected and validated the lifecycle end to end.
Partners & sponsors
Extend
Integration patterns, offer logic, and claim resolution.
Leadership
Priority, escalation path, and the auto-opt-in decision.
Finance & Legal
Cost exposure review and the historical issue.
Product and SKU data had to stay consistent across every placement so Extend could return the right offer. Order and coverage details then had to travel downstream and back through the experience.
Launch included cross-team AHOD meetings with leadership and teams actively testing — and trying to break — the experience in test environments.
We challenged auto-opt-in with a test, not an opinion.
Rather than argue an executive request indefinitely, we measured it — and let the test result decide.
Request
Executives asked for shipping protection auto-opt-in
A top-down request to pre-select protection at checkout.
Concern
The team raised conversion and customer-behavior risk
A real concern, but the data behind it was hypothetical.
Agreed
A two-week A/B test instead of an argument
We could not settle the conversion risk with hypothetical data, so we measured it.
Day 4
~$700K GMS decrease in the treatment
Observed within four days of the test running. This was not annualized.
Escalated
Evidence taken back to senior executives
Ended early
The test was stopped ahead of the two-week mark
Outcome
Auto-opt-in was removed
Impact
09
A connected model, with evidence at each end.
- ~$2.3M
- Historical first-cost exposure identified across seven yearsDiscovered, not represented as recovered
- 100%
- Physical and tangible product coverage for shipping protection
- ~87%
- SKU eligibility for product warranty
- +30%
- Increase in warranty attachment rate for living room and bedroom furniture
Customer & operational outcomes
- Significant decrease in Customer Care contacts for orders delivered outside the 30-day return window.
- A highly manual reporting and claim process became an integrated lifecycle with better visibility.
What the metrics mean
The attachment result is a 30% increase in rate, not 30 percentage points. The historical exposure was identified during discovery; no recovery outcome is claimed.
Reflection
10
The request was to make warranty costs easier to manage. The real problem was the disconnected lifecycle behind them.
Mapping the workflow expanded the product work from supplier tooling into customer experience, financial controls, integrations, reporting, and cross-functional operations. The screen was only where the thread started.