RETURNNESTReturn · Returns disposition & recovery

Every return is trapped cash. Get the most of it back.

ReturnNest grades what comes back, prices every route it could take, checks the claim before any credit goes out, and posts the result through your ERP once a named person approves.

Run the demo Discuss your returns challenge

Stacks of shrink-wrapped cardboard cartons on pallets in an empty office space

ReturnNest at a glance

1Reads

Return claim and original order
Grade from the returns desk
Batch shelf life and stock

2Decides

ReturnNest

Best route per grade
Credit now, or hold for review
Return-to-vendor claims

3Produces

Goods receipt for restocked units
Repack and rework orders
Credit and debit notes

↺ Learns Realized recovery per route, set against the estimate back to 1 · Reads

The business problem

The corner of the warehouse nobody prices

It is quarter end. A distributor sends back 240 cartons of room fragrance and claims the whole lot was damaged in transit. The returns bay is already full. Nobody has time to open every carton, so habit decides: credit the claim and write the lot off, or push it all back into stock.
Either way, money leaks. Intact cartons get scrapped. Crushed cartons that only needed a new outer box get written off. Near-expiry stock goes back on the shelf and expires there. The credit note goes out before anyone checks the carrier's proof of delivery.
Finance sees a write-off line weeks later. Nobody can say what that return was worth, because nobody priced it.

Why the usual fixes fall short

The ERP's returns workflow records the return and moves the stock. It does not price the routes a unit could take.
Spreadsheets can't keep up with grade, shelf life, current demand and supplier terms for every carton, so teams fall back on blanket rules.
Claim evidence sits with sales, credit notes with finance and the goods with the warehouse. Nobody holds all of it at the moment the credit is issued.

Every return keeps getting the quickest decision, and the value it held keeps leaking out as write-offs, expired stock and unchecked credit.

Working capitalReturned stock waits in the bay for weeks before anyone decides, and the cash stays in it.
MarginRecoverable product is written off, and near-expiry stock is restocked only to expire later.
Claims and creditDamage claims are credited without a check against delivery records or the account's claim history.
Audit trailEach return decision is re-keyed into the ERP by hand, with no record of why a route was chosen or who chose it.

In plain terms

What ReturnNest is, and how it fits

What it is: A returns decision layer. For each returned unit or batch it records the grade, prices every route the unit could take, and recommends the one that recovers the most after cost.
Who uses it: Heads of supply chain and finance at manufacturers and distributors with distributor, retail or export returns. Day to day: the returns desk, finance (credit and debit notes, write-offs) and sales operations (distributor claims).
What it does: Grades each return, prices five routes (restock, refurbish, return to vendor, liquidate, scrap), checks the claim before any credit, and drafts the inventory and finance entries for a named approver. A recovery ledger then records what each route actually returned.
How it fits your systems: It sits beside your ERP and warehouse system and adds the decision to the returns process you already run. It reads batch, shelf life, order and supplier terms, and writes back only through your ERP's own documents after approval. Works with SAP, Oracle, Microsoft Dynamics and Tally; connectors are configured per deployment. The physical return movement stays with your transport system or with TransNest.
What is different: The route choice is a calculation you can inspect, with every input shown, so finance can follow it line by line. It runs inside your own network with small local models only: no return record or photo goes to an outside AI service.

See it work

A distributor return at quarter end

A simulated scenario with representative data. Grade a returned lot, see five routes priced, check the claim, and approve the postings as the Finance Controller.

A simulated scenario on representative data. You choose the option and give the approval; the numbers follow your choices.

End-to-end workflow

From data to decision to outcome

What goes in, what ReturnNest works out, what gets decided, what happens next, and how the result comes back.

InputsDATA

Return claim and original order
Grade from the returns desk
Batch shelf life and stock
Demand, price and clearance rates
Supplier return terms and the carrier's proof of delivery

IntelligencePREDICT

Net recovery for each route: expected value minus handling, repack and freight cost
A local claim-check model: serial returners, value anomalies, reason mismatches
An optional camera model on your own servers that proposes a grade for the operator to confirm
Abstains to manual review when grade or demand data is too thin

DecisionsDECIDE

Best route per grade
Credit now, or hold for review
Return-to-vendor claims

ExecutionEXECUTE

Goods receipt for restocked units
Repack and rework orders
Credit and debit notes
Write-off journals
All posted through your ERP, after approval

FeedbackLEARN

Realized recovery per route, set against the estimate
Supplier acceptance of return-to-vendor claims
Claim review outcomes, fed to the claim-check model
Cost settings corrected from actual repack and disposal costs

Modules

Capabilities, organized by the work they do

10 modules in 4 groups. Open a module for its inputs, outputs, users and an example.

Receive and grade3 modules Decide the route and the credit2 modules Post inside your own systems3 modules Prove and learn2 modules

Receive and grade

Know exactly what came back, and in what condition, before anyone decides what to do with it.

Return intake and receiptRecords each return against its original order and batch, with the stated reason and what physically arrived.

Challenge: Returns arrive with a reason code and little else, so the claim, the order and the goods are never in one place.
Inputs: Return authorization, original order and invoice, batch number, quantity, stated reason, receiving scan.
Outputs: One return record linking claim, order, batch and received quantity, with a flag when what arrived differs from what was claimed.
Used by: Returns desk, warehouse receiving.
Value: Every later decision starts from one record instead of three spreadsheets.
Works with: Reads orders and batches from your ERP. The return shipment itself can come from your transport system or TransNest.
In action · illustrative: A distributor claims 240 cartons. The dock scans 240 across 6 pallets, and the record links them to the original invoice and batch FX-0425.

Condition gradingGives every unit or carton a grade from one scan. Batch and remaining shelf life are pulled in automatically, and a guided checklist sets the same standard on every shift.

Challenge: Grading every return by hand is too slow, so it is skipped, and good stock is scrapped along with the bad.
Inputs: Barcode scan, checklist answers (seal, crush, leak, label), batch and shelf life from your ERP or Stocki, optional photos.
Outputs: A grade per unit or carton, signed off by the operator, with its confidence and a flag when the cause of damage is unclear.
Used by: Returns desk operators, quality.
Value: Every unit carries a grade the route pricing can act on, and grades stay consistent across operators and shifts.
Works with: Runs on a handheld scanner app. Batch and shelf life come from your ERP or Stocki, so nobody types them.
In action · illustrative: 24 leaking cartons are graded C. The checklist points to a cap defect from supplier lot CP-77 and flags it for quality to confirm before anyone blames the supplier.

Photo grading on your own serversFor high volumes, a camera model proposes the grade and the visible defects from one photo of each unit. The operator confirms every proposal.

Challenge: Manual grading does not keep up at peak return volumes, and security teams block tools that send product photos to an outside service.
Inputs: One photo per unit or carton, and labeled photos of your past returns to train on.
Outputs: A proposed grade and defect list with a confidence. Photos the model is unsure about go straight to the operator.
Used by: Returns desk operators, quality.
Value: Faster, more consistent grading at peak, with no photo leaving the building.
Works with: Optional. An image-recognition model, not a language model, running on your own server; one modest GPU covers high volumes. The operator confirms every grade.
In action · illustrative: During a post-festival return wave, the operator photographs each crushed carton. The model proposes grade B for 46 of 48; the two it is unsure about are graded by hand.

Decide the route and the credit

Price every route a return could take, and check the claim before any credit goes out.

Net-recovery optimizerPrices every route a unit could take (restock, refurbish, return to vendor, liquidate, scrap) as expected recovery minus cost, and recommends the one worth the most.

Challenge: Blanket scrap throws away recoverable value, and blanket restock puts near-expiry stock back on the shelf to expire there.
Inputs: Grade, remaining shelf life, demand and price, clearance-channel rates, repack, freight, discount and disposal costs, supplier return terms.
Outputs: A recommended route per unit or grade, with its estimate and every input shown. It abstains to manual review when the data is thin.
Used by: Returns desk lead, finance.
Value: More value back from every return, with the reason for each route on record. Same inputs give the same decision, so finance can audit it.
Works with: Shelf life and stock from your ERP or Stocki. Demand from your planning system or PlanningNest. The calculation is deterministic; no generative model is involved.
In action · illustrative: With 70 days of shelf life left, intact cartons are worth most back in stock. At 45 days, the same cartons are worth more in the clearance channel today.

Claim check before creditMatches each damage or return claim to the original order and scores it against the account's history before any credit note is raised.

Challenge: Credit notes are issued on trust, so false damage claims and serial returners go unchecked, sometimes to offset a scheme liability.
Inputs: The claim, the original order and dispatch, the carrier's proof of delivery, the grading result, the account's claim history and the scheme calendar.
Outputs: A claim score with its reasons (was this batch shipped to this account, claim spikes, value outliers, reason mismatches) and a proposed split between credit now and credit held for review.
Used by: Finance, sales operations.
Value: Credit goes out for verified damage. A suspicious claim is held for a person to review, and every hold carries its evidence.
Works with: A local model trained on your own past claims and their outcomes, running on ordinary servers. The model scores; a person decides.
In action · illustrative: A claim on 240 cartons meets a proof of delivery that notes damage on 2 pallets. Credit for the 72 damaged cartons is proposed now; the rest is held for sales operations.

Post inside your own systems

Turn each approved decision into the right ERP entries, inside the returns process you already run, with every record kept in your network.

Posting to inventory and financeTurns each approved decision into the right ERP transactions, with the grade, the evidence and the approval attached.

Challenge: Return decisions are re-keyed by hand, late, error-prone and without a record of why.
Inputs: The approved route and credit decision.
Outputs: Goods receipt (restock), repack order (refurbish), return order and debit note (return to vendor), clearance sales order (liquidate), write-off journal (scrap), credit note to the customer against the original invoice.
Used by: Finance, inventory control.
Value: Inventory and the books reflect the return on the day it is decided, with no re-keying and an audit trail on every entry.
Works with: Posts only through your ERP's own documents. Nothing is written before approval.
In action · illustrative: Once the Finance Controller approves, a goods receipt for 168 cartons, a repack order for 48, a debit note to the cap supplier and a credit note for 72 cartons are posted.

Fits your ERP returns processAdds the decision to the returns workflow you already run. Your ERP keeps the process; ReturnNest decides the outcome and writes it back into the same process.

Challenge: The ERP runs the returns process and records a reason code, but it cannot say which outcome is worth the most.
Inputs: Open return orders and their status from your ERP's returns workflow, reason codes and approval steps.
Outputs: The chosen route, the claim decision and their evidence, written back to the open return order so the next workflow step runs as usual.
Used by: Returns process owners, ERP administrators, finance.
Value: Nothing in the returns process is replaced. Net-recovery routing, claim checks and learning are added on top of it.
Works with: Works with SAP, Oracle, Microsoft Dynamics and Tally returns flows; connectors are configured per deployment.
In action · illustrative: A return order waits at the inspection step in the ERP. ReturnNest adds the grade, the priced route and the claim score; once approved, the ERP moves the order on to repack and credit.

On-premise boundaryKeeps every return record, photo and model inside your network. The route choice is a calculation, and the only models are small local ones.

Challenge: Security teams block tools that send stock records or product photos to an outside AI service.
Inputs: Your servers or private cloud, ERP and Stocki data, the claim-check model and the optional camera model.
Outputs: Route decisions you can reproduce from their inputs, and local model scores, with no call to an outside AI service.
Used by: IT and security, finance.
Value: A security review that starts from a short list of local components, and decisions finance can reproduce.
Works with: On-premise or private cloud. The claim-check model runs on ordinary CPU servers; only high-volume photo grading needs a GPU.
In action · illustrative: The security review lists three components on the company's own servers: the route calculation, the claim-check model and the camera model. No return record or image leaves the site.

Prove and learn

Keep score of what each route actually recovered, and feed it into the next decision.

Recovery ledgerRecords what each route actually recovered, per return, route, SKU and distributor, against the estimate and an agreed baseline, and feeds the gaps back to the optimizer.

Challenge: Nobody can say how much value returns recovered last quarter, so nobody can improve it, and software fees get paid whether or not it improves.
Inputs: Postings, sales of restocked and liquidated units, supplier credits, disposal costs.
Outputs: Recovered value by route, SKU and customer, estimate-versus-actual gaps fed back to the cost settings, and a value statement finance can audit.
Used by: Finance, supply chain leadership.
Value: A recovered-value figure the CFO can audit, sharpened as real outcomes tune the next decision. It is also the basis for value-linked pricing: a fee per return processed plus a share of the recovered value the ledger proves.
Works with: A baseline agreed in writing before measurement starts.
In action · illustrative: The ledger shows repack running at $16 a carton instead of the $14 assumed, and the next estimate uses $16.

Apparel fit-reason packReads apparel return reasons and reviews with a local text model and names the fit problem behind them.

Challenge: "Too small" returns get written off as noise when the real cause is a size curve or a pattern.
Inputs: Return reasons, customer reviews, size, style and fabric lot.
Outputs: A named fit cause by style and size, sent to merchandising and sourcing.
Used by: Merchandising, sourcing, product development.
Value: Fewer repeat returns from the same fit problem.
Works with: Industry pack for apparel. Needs return reasons captured as text; the text model runs locally.
In action · illustrative: Returns citing "tight at the waist" cluster on one style's two largest sizes from a single fabric lot. The pack flags it to the size-curve owner.

Intelligence and automation

What the intelligence actually does

Five levels, from seeing what is happening to learning from it.

Visibility

Every open return in one queue, by grade, days in the returns bay, value at stake and credit claimed.

Prediction

Expected recovery for each route from shelf life, demand and clearance rates, and a claim-check score before any credit.

Decision

Recommends the route with the highest net recovery per grade, and which part of a credit to issue now or hold, with every input shown.

Execution

Drafts goods receipts, repack orders, credit and debit notes and write-offs, posted through your ERP only after approval.

Learning

Compares realized recovery with the estimate for each route and corrects the cost and demand settings; claim reviews feed the claim-check model.

Runs automatically

Pulling batch, shelf life, order and supplier terms for each return
Pricing every route and ranking them
Scoring each claim before credit
Sending a return to manual review when the data is too thin

Always waits for a named person

Every credit note and debit note (Finance Controller)
Every write-off (Finance Controller)
Holding or rejecting a customer's claim (sales operations, with finance)
Every grade proposed by the camera model (returns desk operator)
Any change to route costs, clearance rates or shelf-life rules

Business value

What it helps you work toward, and how we measure it

We agree the baseline with you in writing before a pilot starts. These are the outcomes we measure.

More value recovered per return

Each grade goes to the route worth most after cost, instead of one blanket rule.

Realized recovery per return from the recovery ledger, against a blanket-rule baseline agreed in writing at the start.

Fewer near-expiry write-offs

Short-dated stock goes to clearance while it still has value, instead of back on a shelf.

Units restocked from returns that later expire unsold, by SKU class, before and after.

Credit only for verified damage

Claims are checked against delivery records and history before the credit note goes out.

Share of claimed credit issued after a documented check, and how many held claims are upheld on review.

Faster decisions in the returns bay

Returns stop waiting for someone to find time to decide.

Days from receipt to disposition, from the return record timestamps.

A posting trail you can audit

Every return entry links to its decision, evidence and approver.

Share of return postings linked to a decision record with an approver.

Illustrative example

A distributor returns 240 cartons and claims transit damage. Graded, 168 are intact, 48 have crushed outer cartons and 24 have leaking caps. Written off, the lot recovers nothing and costs about $720 to dispose of. Routed by grade (restock, repack, return to vendor) at 70 days of shelf life, the same lot recovers about $19,000, and the claim check holds credit on the 168 intact cartons for review. Illustrative figures, not a customer result.

Enterprise readiness

Integration, control and deployment

ERP & system integration

Reads orders, batches, shelf life, stock and supplier terms from SAP, Oracle, Microsoft Dynamics or Tally, read-only first, and posts only through the ERP's own documents after approval. Connectors are configured per deployment.

Data sources & connectivity

Return records, photos and claim history stay in your environment. Route pricing is a calculation; the claim-check and camera models are local.

Deployment & implementation

On-premise or private cloud, with no outside AI service in the loop. Rollout starts in shadow mode next to the current returns process, before anything is posted.

Configuration & business rules

Routes, handling and repack costs, clearance rates, minimum shelf-life rules and supplier return terms are settings, not code.

Roles, access & approvals

Returns desk (receive and grade), finance (credit, debit, write-off), sales operations (claim review) and administrators for cost and routing settings.

Security, governance & audit

Single sign-on, role-based access and an audit log of every grade, override and approval.

Human oversight & exceptions

Every value-bearing entry waits for a named approver, the system abstains when data is thin, and model changes must pass a test set of past returns before release.

Scale & ongoing optimization

Works per unit or per batch, across more than one returns center, with the claim-check model running on ordinary servers.

Straight answers

Questions about ReturnNest

Something else? Ask a specialist.

Do we have to replace our ERP returns process?

No. Your ERP keeps the returns workflow, the stock and the books. ReturnNest adds the grade, the priced route and the claim check, and writes the approved decision back into the same process through the ERP's own documents.

How does ReturnNest work with TransNest?

TransNest moves the goods: it handles return shipments, records the grade a person enters and routes the goods to the right depot. ReturnNest decides what the goods are worth and where they should go: it prices the routes, checks the claim and drafts the postings. Each works without the other.

Will it post entries to our ERP by itself?

No. It drafts goods receipts, credit and debit notes and write-offs, and posts them through your ERP's own documents only after a named approver signs. Until then it only reads.

Does any of our data go to an outside AI service?

No. Route pricing is a calculation, and the claim-check model and the optional camera model run inside your network.

How do we start, and how is it priced?

With a read-only sample of past returns and the real cost of each route. We shape the grading checklist and approval rules with your team and run in shadow mode next to the current process before anything is posted. Pricing can follow the recovery ledger: a fee per return processed plus a share of the recovered value it proves, against a baseline agreed in writing first.

Next step

Discuss your returns challenge

Returns are where a supply chain's leaks become visible: write-offs, expired stock and unchecked credit. ReturnNest turns each one into a priced, approved, recorded decision, and the recovery ledger shows what it was worth.

Talk to a specialist

For manufacturers and distributors with steady distributor, retail or export returns, an ERP that records returns without pricing them, and a finance team that wants every credit backed by evidence.

In the same loop