Feature Deep-Dive ·Transportation Management

Returnable Packaging Intelligence (RPM)

Asset-loop AI for beverage, automotive, and industrial supply chains — every keg, crate, and pallet accounted for.

✦  Fully automated route settlement lifted from 70% to 90% at a global alco-bev leader — and 28% cut in LSP excessive-stay payments.

The problem it solves

A beverage distributor sends out 120 kegs with a single route. Forty-three come back. Eighteen come back chipped. Nine never come back at all, and nobody will know for another 45 days because the reconciliation happens manually in Excel against a rate card that changed two months ago. The retailer was charged a deposit for every keg that left the warehouse; three weeks later the credit note is wrong; the driver’s LSP settlement sheet shows an excessive-stay charge the customer disputes; a quarter later the finance team writes off the whole mess as “shrinkage” and moves on. Multiply this by 2 million trips a year and the returnable-asset loop becomes a nine-figure leak that nobody owns — part logistics, part finance, part inventory, fully invisible. This is true for kegs, for crates, for IBCs, for pallets, for gas cylinders, for automotive totes, for beer tub returns from a festival on a Sunday morning. The asset is always worth something. The loop is almost never closed cleanly.

What it is

Returnable Packaging Intelligence (RPM) is Shipsy’s asset-loop brain — the system that tracks every returnable unit from issue to return, reconciles it against trip and customer-level expectations, flags aging and loss before they become writeoffs, and settles the commercial loop (deposit, credit, LSP stay-payment, damage deduction) without a human in the spreadsheet. It is not an asset-tracking module; it is a closed-loop settlement system that runs inside Shipsy’s TMS and talks to the ERP stock ledger, the primary-distribution planner, and the dispute agent Vera. What’s new-age about it: the reconciliation isn’t batch — it’s event-driven. The moment a truck leaves the brewery with 200 crates, the expected return window opens. The moment the truck gets back with 187, the 13-crate exception is already a case with photo evidence, driver statement, and customer confirmation — not a line in a monthly writeoff report.

Core capabilities

Capability What it does
Issue & return ledger Every returnable asset — keg, crate, IBC, pallet, tote, cylinder — carries a scan identity. Issues are bound to trips and customers; returns are bound to arrivals. The ledger is the single source of truth that the ERP, finance, and ops see together.
Automated route settlement At trip close, the system reconciles issued vs returned assets, expected vs actual stay-time at each stop, expected vs actual route completion — and posts the settled result to the ERP. Automation coverage climbs from 70% at baseline to 90%+ in production.
Aging thresholds & loss prevention Each asset class carries an age threshold (e.g., “a steel keg should cycle back within 21 days”). The system watches the ledger and flags anomalies — assets out past threshold, customers with rising non-return rates, LSPs whose round-trip cycle is drifting.
LSP excessive-stay detection Stay-time at each customer is measured from arrival scan to departure scan. Excessive-stay payments to LSPs — a major source of silent leakage in beverage distribution — are flagged before they’re paid, with evidence bundled. In production: 28% reduction.
Damage assessment at return Returns are scanned with a guided photo-capture flow; the damage classifier scores condition, deducts per rate card, and generates the credit note in the same workflow. No second-pass inspection queue.
Deposit & credit reconciliation Customer-level deposit ledger syncs to the ERP in near real time. Credits issue automatically when returns reconcile; disputes escalate to Vera with the full evidence chain.
Predictive loss model Combines customer history, LSP history, route characteristics, seasonal patterns, and asset class to predict which issues are most at risk of not returning. Used to prioritize recovery workflows and adjust deposits dynamically where allowed.
Multi-asset configuration One install handles kegs, crates, IBCs, totes, pallets, cylinders, reusable tubs, automotive packaging, and retail returnable display units. Each class carries its own rate, age threshold, damage schema, and settlement logic.
Driver & LSP mobile flows The Shipsy driver app drives issue, return, and damage scans inline with pickup/delivery — not as a side task. The LSP supervisor sees the same ledger on a mobile view for exception handling.
Integration with Vera Disputes — whether customer-initiated (“we returned those crates”) or LSP-initiated (“the stay was legitimate”) — land with Vera, which autonomously resolves against contract, stock ledger, and GPS evidence.
ERP stock ledger sync Bidirectional sync to Oracle, SAP, Microsoft Dynamics, and custom inventory stacks. The returnable ledger never drifts from the stock ledger because it writes back on every event, not nightly.
Compliance & audit trail Every issue, return, damage assessment, settlement entry, and dispute is immutable and attributable. Auditors trace a deposit credit back to a photo, a GPS trace, and a timestamped scan in under 30 seconds.

How it works

The system models a returnable as a first-class entity with a lifecycle — issued, in transit, at customer, returning, received, reconciled, settled — and attaches the commercial impact (deposit, rate, damage schedule) to that lifecycle. Every scan, every stop, every ERP post becomes a state transition. The reconciliation engine runs on the event stream, not on a nightly batch: the moment a return scan hits the event bus, the matching issue is debited, the expected-return window is closed, any remainder becomes an exception, and the commercial posting is queued for the ERP. The predictive loss model runs on the same event history, flagging customers and LSPs whose behaviour is drifting before the drift becomes a quarter-end surprise.

graph TB A[Driver App Issue Scan] --> E[RPM Event Stream] B[Return Scan & Damage Photos] --> E C[LSP Stay-Time Sensors] --> E D[Customer Confirmations] --> E E --> F[Reconciliation Engine] E --> G[Aging & Loss Predictor] F --> H[Settlement Generator] G --> I[Exception Queue] H --> J[ERP Stock & Finance] I --> K[Vera Dispute Agent] H --> L[LSP Payment Rail] I --> M[Ops Recovery Workflow]

The workflow below traces a single returnable cycle — a trip with 200 crates, a 24-hour round trip, and a 13-crate shortfall — showing how the loop closes itself without a spreadsheet and without a manual chase. In production, 90%+ of trips settle fully automatically; only the true exceptions surface as cases.

sequenceDiagram participant D as Driver participant R as RPM Engine participant E as ERP participant V as Vera participant L as LSP Finance D->>R: Issue scan — 200 crates to Route A R->>E: Debit customer deposit ledger D->>R: Stop-level scans + stay-time D->>R: Return scan — 187 crates + 4 damaged R->>R: Reconcile, assess damage R->>E: Credit note with damage deduction R->>V: 13-crate exception → dispute case R->>L: Stay-payment settled (or flagged) V-->>L: Resolution posted

Proven outcomes

Customer type & scale Outcome
A global alco-bev leader operating across 70+ countries, 2M+ annual distribution trips Fully automated route settlement lifted from 70% to 90%; 28% reduction in LSP excessive-stay payments; $25M+ in carrier/vendor disputes autonomously resolved (with Vera)
A major beverage bottling group Returnable-asset loss materially reduced; deposit reconciliation cycle compressed from weekly batch to event-driven settlement
A global 3PL with European roots Cross-customer returnable-asset visibility enabled across shared routes; LSP stay-payment leakage shut down
A MENA retail conglomerate with 80+ years operating multi-brand portfolios Returnable retail display and tote loop instrumented across multi-brand distribution; shrinkage on reusable assets materially reduced

Integrations

  • ERP & finance — Oracle ERP Cloud, SAP S/4HANA, Microsoft Dynamics 365, NetSuite; bidirectional stock and deposit ledger sync.
  • WMS — Manhattan, Blue Yonder, SAP EWM; issue and return events mirrored to warehouse inventory.
  • Driver & LSP apps — Native Shipsy driver app; LSP supervisor mobile view; partner SDKs for embedded scanning.
  • Telematics — Wialon, Wheelseye, Samsara and native OEM telematics for stay-time and GPS evidence.
  • Dispute automation — Vera for autonomous dispute resolution against contract terms and evidence chain.
  • Customer systems — EDI 214/990 flows; retailer deposit-ledger connectors for major grocery and HoReCa accounts.
  • Data & analytics — Snowflake, BigQuery, Databricks connectors; standard BI stack for finance-grade reporting.
  • Regulatory — GDP and traceability modes for pharma returnables; bonded-movement flags for alco-bev where required.

Deployment

Most enterprises go live with RPM in 10-14 weeks, with a pilot distribution circuit live by week 5-7.

  • Phase 1 · Discovery (Week 1-3) — Baseline the returnable-asset loop: which assets, which routes, which customers, which LSPs, current loss rate, current settlement-automation rate. Build the asset taxonomy.
  • Phase 2 · Configuration (Week 3-7) — Model asset classes, rate cards, damage schedules, age thresholds, LSP contracts. Wire ERP stock ledger. Stand up scanner flows for drivers and LSP supervisors.
  • Phase 3 · Pilot (Week 7-10) — One distribution circuit goes live end-to-end. Success criteria: settlement-automation rate +15 points from baseline, LSP excessive-stay payments down 20%+, aging breaches detected within SLA.
  • Phase 4 · Scale (Week 10-14+) — Roll by region. Predictive loss model activates at 60-90 days of clean event data. Vera is wired for autonomous dispute resolution at scale.

Governance runs through a Returnable Asset Council — distribution ops, finance, procurement, LSP partner management — that reviews aging thresholds, damage schedules, and rate cards quarterly. Changes to commercial logic flow through a documented approval path because they hit customer-facing credit notes.

Security & compliance

  • SOC 2 Type II, ISO 27001, and GDPR-aligned by default.
  • 21 CFR Part 11, GDP, and GMP modes available for pharma-returnable flows.
  • Immutable audit trail on every issue, return, damage call, settlement, and dispute. Evidence (GPS trace, photos, scans) is retained per regional audit and customs requirements.
  • Role-based access across driver, LSP supervisor, distribution ops, finance, and audit personas.
  • Three-tier confidence scoring on the damage classifier and the predictive loss model — the bottom tier flags for human review, never auto-posts a deduction.
  • Human-in-the-loop for rate-card overrides, LSP contractual disputes, and high-value writeoffs.

Case study callouts

Global alco-bev leader · 70+ countries · 2M+ annual distribution trips

“Moved fully automated route settlement from 70% to 90%. Cut LSP payments for excessive stays at customer location by 28%. Reduced failed deliveries due to ‘customer not present’ by 50%. Vera — wired into the returnable-asset loop — autonomously resolved $25M+ in carrier and vendor disputes.”

Read the full case study

A major beverage bottling group

“Compressed returnable-asset reconciliation from weekly batch to event-driven settlement. Deposit ledger now mirrors the ERP stock ledger in near real time; customer credit cycles shortened; asset-loss rate materially reduced.”

Read a related beverage deep-dive

A global 3PL with European roots

“Instrumented cross-customer returnable-asset visibility across shared routes. LSP stay-payment leakage shut down — the stay-time evidence chain made the disputes self-resolve.”

Read the 3PL perspective

Frequently Asked Questions

How long does deployment typically take?

Most enterprises run a pilot distribution circuit in 5-7 weeks and full regional scale in 10-14 weeks. The first region is heaviest — asset taxonomy, rate cards, and ERP integration are the long pole. Subsequent regions clone the template and typically go live in 3-4 weeks each.

Do we need RFID on every asset?

No. Most customers run with barcode scan identities on the asset or the cage that holds the asset. RFID is supported where asset class and handling economics justify it — typically steel kegs and high-value industrial totes. Vision-based scanning is also supported for reusable display units.

How does RPM work with our existing ERP?

Bidirectional. RPM writes issue, return, damage, and settlement events into the ERP stock and finance ledgers in near real time — and reads rate cards and customer master from the ERP. Certified connectors exist for Oracle, SAP, Dynamics, and NetSuite; custom inventory stacks use Shipsy's integration framework.

What's the relationship between RPM and Vera?

RPM handles the loop; Vera handles the disputes that fall out of the loop. When a settlement exception or a customer/LSP dispute surfaces, RPM routes it to Vera with the full evidence chain. Vera autonomously resolves against contract terms — the proof point is $25M+ resolved at a global alco-bev leader.

Can we dynamically adjust customer deposits based on return behaviour?

Yes, where your commercial model allows it. The predictive loss model scores each customer and LSP; the system can flag high-risk accounts for deposit uplift, tiered credit terms, or route-level recovery workflows. In regulated markets where deposits are fixed, the model powers recovery prioritization instead.

How does this work in bonded and regulated alco-bev movements?

RPM carries bonded-movement flags on the asset and trip entities. Customs and excise evidence attaches to the same event chain that drives reconciliation. Many regulated regions (EU, UK, select APAC) are pre-mapped.

We have assets owned by multiple entities — brewery, distributor, retailer. Can the system handle that?

Yes. Each asset carries an ownership attribute that drives the deposit ledger, the rate card, and the settlement logic. Multi-entity reconciliation is a supported flow — particularly important for 3PLs who move pooled assets across clients.

What happens to our existing manual reconciliation team?

Most customers repoint the team to exception management — the 10% of trips that don't auto-settle — and to predictive loss recovery. The work shifts from "reconcile 100% manually" to "resolve 10% with evidence in hand." Headcount is redeployed, not cut.