Returns Optimization in 2026: Optoro, Happy Returns, Loop, ReturnGo, and the Keep-It Decision
The returns vendor landscape sorted honestly — what Loop does that Returnly never did, where Optoro's reverse-logistics platform actually fits, and how AI-driven keep-vs-return decisions reshape the unit economics.
Returns are the unglamorous half of e-commerce. The forward supply chain — pick, pack, ship — is heavily optimised and well-measured. The reverse supply chain — return authorised, package shipped back, inspected, restocked or liquidated — is where the margin quietly evaporates. By 2026 the aggregate US return rate sits in the high teens for online orders, with apparel routinely above 30 percent, and the reverse logistics cost per return has become a meaningful percentage of order value. The vendor landscape that grew up to solve this is a mix of consolidation (PayPal’s Happy Returns acquisition, Affirm’s Returnly acquisition), a few well-funded specialists (Loop, ReturnGo), and the old reverse-logistics incumbents (Optoro).
This is where returns AI actually sits.
The economics that drive everything#
A typical return on a 50 USD apparel item costs the retailer somewhere between 8 and 15 USD in shipping, inspection, repackaging, and processing once the item arrives back at the warehouse — and that’s before considering the margin loss from items that come back damaged, used, or out-of-season. Free returns trained two decades of consumer behaviour around bracket buying (order multiple sizes, keep one), wardrobing (wear once, return), and the increasingly visible category of return fraud (empty packages, switched items, false damage claims).
The 2024-2025 retailer response has split in two directions. Some — Zara, H&M, Boohoo, Uniqlo, J.Crew, Abercrombie — introduced return fees, with measurable reductions in return rate but a notable customer-acquisition cost on social channels. Others — Walmart, Amazon, Costco — kept free returns but invested heavily in AI-driven returns intelligence to reduce the cost-per-return through smarter handling, fraud detection, and selective keep-it decisioning.
The vendor space sits on both sides of that bet.
Optoro#
Optoro (Washington DC-based, founded 2010, secured a 100-million-dollar round in late 2024 after a difficult 2022-2023 stretch) is the incumbent reverse-logistics specialist. The product covers the post-return part of the chain — inspection, grading, disposition (restock to warehouse, route to liquidation, donate, recycle), and the Bulq and BlinqStore secondary sales channels.
Optoro’s strength is the disposition optimisation — for a returned item, where should it go to maximise recovery value net of handling cost? The AI does the math: restock if the item passes inspection and is in-season, route to a B-stock channel if used but functional, liquidate to a bulk buyer if not worth individual handling, donate or recycle as the floor. Best Buy, Target, and a long list of large retailers have run Optoro through this disposition workflow for years.
Where Optoro is less competitive: the consumer-facing return experience. That’s where Loop, ReturnGo, and Happy Returns sell.
Happy Returns (PayPal)#
Happy Returns (acquired by PayPal in 2021) built the “drop off your return at a nearby retailer” model — over 12,000 Return Bar locations across the US as of late 2025, including FedEx Office, Staples, Petco, and a long tail of independent retailers. The wedge is the consumer experience: no box, no label, scan a QR code at a Return Bar, walk out. The retailer benefits from aggregated pickup (Happy Returns batches returns at the bars and ships consolidated loads, materially cheaper than individual returns).
PayPal’s strategic review in 2024-2025 raised public questions about the future of Happy Returns inside the PayPal portfolio, and the rumoured spin-out was discussed in trade press through Q4 2025 without resolution. Operationally, the platform continues to work and the merchant base continues to grow. The 2026 question is whether PayPal commits to the category or divests.
Returnly (Affirm) and the wind-down#
Returnly was the Shopify-native returns startup acquired by Affirm in 2021 for around 300 million USD. The integration with Affirm never produced the cross-sell story the acquisition assumed, and Affirm wound down the standalone Returnly brand through 2023, folding the product into the broader Affirm merchant platform. By 2026 most former Returnly customers have migrated to Loop, ReturnGo, or the native Shopify returns flow. The lesson the category took from Returnly: returns are operationally adjacent to payments but commercially distinct, and the acquired-into-fintech model didn’t hold.
Loop Returns#
Loop (Columbus Ohio-based, valued at roughly 250 million USD in its 2021 round) is the leading Shopify-native returns platform in 2026. The wedge is workflow — Loop converts returns into exchanges, store credit, or upsells at the moment of return initiation, with measurable retention impact. Brooklinen, Princess Polly, Allbirds, and a deep bench of direct-to-consumer Shopify brands run Loop, and the 2024-2025 product investment added meaningful AI — return-reason classification, item-condition prediction from customer-uploaded photos, and dynamic policy enforcement based on customer history.
Loop’s strength is the commerce integration — the returns flow becomes a retention surface rather than a cost center. The constraint is the Shopify ecosystem; Loop on other commerce platforms is a smaller story.
ReturnGo#
ReturnGo (Israeli-origin, US-expanded) is the closest competitor to Loop, with a similar Shopify-native posture and a slightly different bet — heavier on the AI-driven recommendation of alternatives at the moment of return. “We see you’re returning this — here’s a different size in stock, here’s a different colour, here’s store credit with a 20 percent bonus.” The conversion of returns into exchanges or alternative purchases is the measurable revenue impact.
ReturnGo’s customer base skews to mid-market Shopify brands; the platform has been quietly winning competitive deals against Loop on workflow flexibility and pricing.
The AI-driven keep-vs-return decision#
The most significant 2024-2026 development in returns AI is the keep-it decision. For low-value items or items the retailer cannot economically restock, the AI calculates the disposition cost (return shipping plus inspection plus likely resale-channel value) and offers the customer a refund without requiring the return. Amazon shipped this at scale in 2023; Walmart and Target followed; a long tail of smaller retailers integrated the pattern through Optoro, Loop, and the cloud-native returns APIs.
The economic logic is straightforward — for an item where the round-trip return cost exceeds the recoverable value, the optimal disposition is to refund the customer and not ship the item back. The AI risk is fraud — customers who learn the keep-it threshold and game it — and the operational risk is brand perception (“Amazon just lets people keep stuff for free”). Production implementations vet the customer history, product category, and item condition before triggering keep-it, and tune the threshold over time.
The keep-it pattern is the closest returns AI gets to genuine generative-AI utility — the decision combines a regression on disposition cost, a classifier on fraud risk, and increasingly an LLM-generated customer-facing explanation. For a typical retailer the keep-it decision applies to a meaningful single-digit percentage of returns, and the cost savings net of incremental fraud are real.
Returns abuse detection#
The other meaningful AI category in returns is abuse detection — identifying customers who systematically game the returns policy through bracket buying, wardrobing, empty-box returns, switched-item returns, and serial damage claims. Riskified, Signifyd, and Forter — primarily fraud-prevention vendors — extended into returns abuse over 2023-2025. Loop and ReturnGo built native abuse-detection layers. Amazon and Walmart run internal models that quietly flag and friction-up abusive customers.
The civil-liberties tension is real — banning a customer for excessive returns produces social media incidents, and the line between legitimate consumer behaviour and abuse is genuinely fuzzy. The vendors that handle this well lean toward friction (require photos, require warehouse inspection before refund, hold refund until inspection) rather than ban, and they make the rules transparent to the customer.
What we recommend in 2026#
For a retailer scoping a returns program:
- Shopify-native direct-to-consumer brand — Loop or ReturnGo as the consumer-facing returns workflow, with the AI exchange-conversion layer turned on from day one.
- Larger multi-channel retailer with existing reverse-logistics footprint — Optoro for the disposition side, plus Happy Returns or equivalent for the consumer pickup experience.
- Keep-it decisioning — pilot at low-value categories first, measure fraud incidence over 90 days, then expand. Do not roll out across the catalogue without the measurement layer.
- Returns abuse detection — friction-up before ban-out, and make the policy transparent. The customer-experience cost of a wrong ban is meaningfully larger than the abuse loss prevented.
- Returns analytics as the under-invested category — most retailers do not have a clean view of return rate by SKU, customer cohort, channel, and reason, and that gap is the binding constraint on smarter merchandising and sizing decisions upstream.
Related reading#
Where pdpspectra fits#
We help retailers stand up the returns data layer — order, return, inspection, disposition, and customer-history pipelines that feed both the consumer-facing returns platform and the internal merchandising and fraud teams. Our business automation practice covers the workflow side.
Returns are where margin quietly disappears. If you are evaluating Loop, ReturnGo, Optoro, or scoping a keep-it pilot, tell us about the volume.