E-commerce

AI Chatbot for international returns: country rules, documents, and tracking

AI Chatbot for international returns: country rules, documents, and tracking

August 17, 2026

"I am returning from Canada, the portal does not offer a label." "What customs documents for a UK return?" "Will my refund be in euros or dollars?" Three return_intl questions that the domestic return bot (#10) cannot handle.

Loop Returns estimates that 35 to 45% of international return tickets relate to labels, customs documents, or refund delays, cases that can be resolved via self-service if the bot knows the policy by country (Loop Returns, international returns 2026). Narvar observes that cross-border returns generate 2 to 3 times more CS contacts than domestic returns (Narvar, returns 2025).

This guide #380 covers the AI chatbot for international e-commerce returns: country rules, customs documents, transit tracking, and INT-RET-BOT intents. Distinct from intl returns support (#379) (INT-RET agents framework, macros) and return bot (#10) (domestic RMA): here, AI cross-border automation return_intl_*.

Summary

Why automate international returns via AI chatbot?

An e-commerce international returns chatbot answers market eligibility, international label portal links, CN22/23 customs documents, transit times, and currency refunds without an agent ticket, while handing off customs holds to INT-RET (#379).

Five frustrations without a return_intl bot

  • Domestic-only portal: Canadian client does not see the international flow

  • Vague customs documents: return commercial invoice cannot be found

  • Opaque refund timeline: 4 weeks vs 10 days domestically

  • Confusing refund currency: CAD paid, EUR refunded

  • Customs hold without guidance: package blocked, customer panics

Ecommerce Times points out that the transport cost alone for a cross-border return can reach $18 to $40 per unit, making self-service critical to avoid return abandonment (Ecommerce Times, cross-border 2026).

Angle #380

The INT-RET (#379) guide covers agent macros and customer service ops. The returns bot (#10) = domestic exchange/refund. The #378 RET-HELP feeds the domestic corpus. The #380 deploys the AI bot intents return_intl_eligibility, return_intl_label, return_intl_customs, return_intl_tracking.

DTC Example

EU skincare brand shipping to US/UK/CA, 48 return_intl tickets/month before INT-RET-BOT. After eligibility bot + label link + customs PDF + transit time: return_intl_bot_resolution 62%, return_intl_ticket_deflection -41%, customs_hold_escalation quality +35% (documented handoff).

Intl bot vs domestic bot

Bot #10 = 30-day eligibility, FR Loop portal. INT-RET-BOT = shipping country vs return country, CN22/23, 15-25 day transit, FX refund. Separate NLP intents mandatory.

Journey stage

Post-delivery international (label), D+7 stale return tracking, D+14 pending refund, customs hold alert. Order-market contextual bot.

INT-RET Handoff

Customs hold > 7 days, lost return shipment, refund exception → #379 agent. Bot never refunds before warehouse receipt.

ROI of the return_intl bot

Return_intl FAQ ticket = 18 min INT-RET agent. Bot label + customs doc = 90 s. Out of 35 tickets deflected/month: 630 min saved.

Common mistake

Routing a domestic return intent to the international flow or vice versa: UK customer receives a domestic 5-10 day delivery timeline instead of 20-35 cross-border days. Order market detection mandatory.

How does it differ from #379 INT-RET and the return bot #10?

Eight neighboring contents, eight distinct roles in international returns and AI automation.

Intl returns support (#379)

INT-RET (#379): agent framework IR-1 to IR-8, INT-RET-* macros. The #380 = upstream self-service bot, handoff if customs hold or exception occurs.

Domestic returns bot (#10)

Returns bot (#10): exchange, FR refund. Handoff return_intl intents → INT-RET-BOT #380.

Returns help page (#378)

RET-HELP (#378): domestic hub /returns. The #380 indexes /returns-international into RAG corpus.

Shopify Markets Bot

Markets Bot: general language/currency market support. Handoff return_intl_* → INT-RET-BOT.

International support (#100)

International support (#100): global view. The #380 = cross-border returns bot only.

Import customs (#65)

Customs (#65): purchase DDP/DDU. The #380 = export return customs documents bot.

AI Governance (#142)

Governance (#142): bot explains policy, does not refund prior to receipt.

Promise #380

INT-RET-BOT framework, 10 return_intl_* intents, country RAG corpus, market-aware flows, KPI return_intl_bot_resolution.

Which return_intl_* intents should be mapped for the MVP?

Map the return_intl bot MVP intents before RAG corpus and Loop integration.

Ten MVP return_intl intents

  1. return_intl_eligibility: can I return from [country]?

  2. return_intl_label: where to get intl return label?

  3. return_intl_customs_doc: which customs documents CN22/23?

  4. return_intl_cost: free return or high shipping cost?

  5. return_intl_delay: transit delay + refund timeline

  6. return_intl_refund_fx: refund in which currency?

  7. return_intl_tracking: where is my return parcel?

  8. return_intl_customs_hold: parcel blocked at return customs

  9. return_intl_hub: address of local return hub [country]

  10. return_intl_keep_item: prohibitive shipping cost, alternatives

MVP Phase 1 Prioritization

Ship first: eligibility, label, customs_doc, delay. Phase 2: tracking API, customs_hold handoff, keep_item partial.

90-day ticket mining

Export helpdesk return_intl tags. Quantify volume per intent. Top 4 = 70% of bot-resolvable tickets.

NLP Distinction

"How to return?" without country → bot asks for return country. "Return from UK" → route directly to INT-RET-BOT.

Customer verbatims

Index 40 real US/UK/CA ticket phrases for training intent classifier and QA testing.

Intent deny market

return_intl_eligibility false: bot INT-RET-DENY RAG + keep item alternatives if policy permits.

How to structure the INT-RET-BOT framework in eight steps?

The INT-RET-BOT framework structures each international return conversation into eight reproducible bot steps.

Eight INT-RET-BOT steps

  1. IB-1 Detect market: order ship country, customer return country session or question

  2. IB-2 Eligibility check: policy RAG zone EU/UK/US/APAC

  3. IB-3 Mode retour: prepaid label, local hub, keep item partial

  4. IB-4 Documents customs: PDF link CN22/23 + commercial invoice template

  5. IB-5 Portal deep link: Loop intl return URL market-aware

  6. IB-6 Timeline communicate: transit 15-25 days + refund 5-10 days post-receipt

  7. IB-7 Tracking status: carrier API if RMA exists, otherwise explain next step

  8. IB-8 Handoff ou close: customs hold → pre-filled INT-RET agent ticket

Decision tree IB-3

If shipping cost estimate > 50 % product value → propose keep_item 30-40 % partial refund RAG. Otherwise prepaid label flow IB-5.

Personnalisation order

Bot reads Shopify order: ship_country, currency, line items restrictions (cosmetics, lithium). Adapts eligibility and deny messages.

Multilingue markets

Responses EN US/UK/CA. FR EU. Link to multilingual knowledge base for corpus by locale.

Anti-hallucination IB-2

Eligibility never invented: always cites RAG zone block or handoff if market is unknown. Test 20 orders edge cases before go-live.

Session context

Intl post-order confirmation widget: bot pre-loads IB-1 ship country, reducing friction from "from which country are you returning?"

Which country RAG corpus and customs documents should be indexed?

The international returns RAG corpus indexes country rules, customs documents, and transit times by Shopify Markets zone.

Eight INT-RAG blocks

  • INT-RAG-EU: intra-EU prepaid, no customs, refund 5-10 days

  • INT-RAG-UK: post-Brexit CN22/CN23, +7-10 days transit

  • INT-RAG-US: UPS/DHL return center, transit times 10-15 days

  • INT-RAG-CA: non-refundable purchase duties, prepaid label

  • INT-RAG-APAC: high cost, Sydney hub, keep item policy

  • INT-RAG-CUSTOMS: return commercial invoice, HS code, declared value

  • INT-RAG-FX: refund in same currency as order, standard bank variance

  • INT-RAG-DENY: excluded markets, products banned from intl return

Ops policy sync

Corpus updated when INT-RET Notion policy changes. Support ops owner. Monthly review post-new Shopify market.

Customs PDF embed

Stable link /assets/customs-return-checklist.pdf indexed in INT-RAG-CUSTOMS. Bot return_intl_customs_doc sends PDF + 3-step instructions.

RAG Chunking

One chunk per zone + one global customs chunk. Avoid mixing UK rules in US customer answers.

Metadata filters

RAG query filters ship_country + return_country tags. Reduces cross-zone hallucination.

Page /retours-international source

Content from #379 intl page and RET-HELP extension = primary source for bot indexing.

Which conversational flows should be deployed per market?

The return_intl conversational flows vary by market and intent to maximize bot resolution.

return_intl_label Flow

Customer: « return label from Canada ». Bot IB-1 detect order ship US → IB-2 eligible CA → IB-5 deep link Loop intl + IB-4 customs PDF → IB-6 delay 20-30 days total.

return_intl_customs_hold Flow

Customer: « customs held my package ». Bot asks for tracking → quotes INT-RAG-CUSTOMS missing documents → if hold > 7 days IB-8 handoff INT-RET customs_status ticket + tracking.

return_intl_refund_fx Flow

Customer: « refund in dollars or euros? ». Bot reads order currency CAD → INT-RAG-FX same currency → IB-6 refund post-receipt 5-10 business days after transit.

return_intl_cost high Flow

APAC Customer: « return costs 45 USD ». Bot IB-3 keep_item 40 % partial vs label → customer chooses → bot confirms policy, no auto refund (handoff if accepted).

return_intl_tracking Flow

If RMA + Loop API tracking: live status in_transit, customs_clearance, delivered. Otherwise explain IB-6 steps without inventing a date.

Domestic Fallback

Return country = ship country EU intra → redirect to domestic flow bot #10 or RET-HELP link. Do not mix with INT-RET-BOT.

Proactive post-RMA email

Klaviyo intl return trigger: bot widget « Questions about customs documents? » with RMA ID context.

What is the integration for Loop, Shopify Markets, and tracking?

The INT-RET-BOT technical integration connects Loop international returns, Shopify Markets, and carriers tracking.

Loop international API

Deep link return portal by order ID + market. GET return status, tracking, customs hold flag if available. Cache 10 min per RMA.

Shopify order context

Storefront or Admin API: shipping_address.country, presentment_currency, line_items. Real-time bot eligibility without asking for order number if session is logged.

Shopify Markets routing

Market detect from order → select INT-RAG zone block. Align Bot Markets handoff return_intl.

Carrier tracking webhook

DHL/UPS customs hold event → optional proactive bot message "Your return is in customs check, here are the documents".

Helpdesk handoff ticket

Zendesk/Gorgias: pre-filled ticket ship_country, return_country, tracking, customs_status, bot transcript IB-1 to IB-7.

Widget placement

Intl order status page, /returns-international, post-delivery email footer. No active checkout (payment distraction).

ReturnGO / AfterShip

Adapt IB-5 deep link according to installed return app. Document equivalent API mapping for Loop.

What governance: explaining without early refund?

The international returns bot governance defines limits: explain policy and documents, no early refund or customs override.

Bot can say

  • Market eligibility, label portal link, customs PDF

  • Transit time + refund timeline upper range

  • Refund in the same currency as the order policy

  • Keep item options if shipping costs are high policy

  • Tracking status if API is available

Bot cannot do

  • Refund before warehouse receipt (→ #379 agent exception)

  • Release package from customs directly

  • Promise refund timeline < policy minimum

  • Generate label outside Loop API portal

  • Refund purchase import duties (#65 scope)

Align governance (#142) and limits (#124).

Intl bot audit logs

Intent, RAG zone cited, handoff yes/no, order_id hash. Monthly review of 15 customs_hold conversations.

Cross-border GDPR

Do not expose third-party customs invoice data. Customer only sees their own order.

Which return_intl_bot KPIs should be measured?

The international returns bot KPIs measure ticket deflection and customs handoff quality.

Eight key metrics

  • return_intl_bot_resolution: resolved without agent / sessions return_intl intent

  • return_intl_ticket_deflection: decrease in tickets #379 post-bot

  • return_intl_bot_csat: post-conversation survey intl return

  • return_intl_portal_click_from_bot: Loop deep link clicks

  • customs_doc_download_bot: CN22 PDF from bot

  • return_intl_handoff_rate: handoff #379 / sessions intl

  • return_intl_abandon_post_bot: RMA created post-bot, package never shipped

  • rag_miss_return_intl: unanswered questions corpus zone

DTC Benchmark

Target return_intl_bot_resolution > 60%, return_intl_ticket_deflection -35%, return_intl_handoff_rate < 25%.

Dashboard by market

US, UK, CA, EU: volume of intents, resolution, handoff, portal clicks. Share #ops + #support weekly.

Customs_hold Correlation

customs_doc_download_bot vs customs_hold_return_rate: measure impact of proactive PDF.

Rag_miss spike Alert

rag_miss_return_intl +40% week: new market without RAG block or unsynced policy change.

What edge cases and custom hold handoffs should be anticipated?

Seven intl return bot edge cases require handoff or specific INT-RET-BOT rules.

Customer moved post-purchase

Ship FR, return US: bot re-checks IB-2 eligibility return from US, new label origin. Handoff if policy is ambiguous.

Intl gift return without invoice

Bot INT-RAG-CUSTOMS gift return template SKU value. No handoff if PDF is sufficient.

Cosmetics / lithium restrictions

INT-RAG-DENY carrier restriction → keep item destroy policy or handoff #379.

DDP duties purchase question

Redirect intent customs purchase → #65. Bot #380 = export return only.

Marketplace order intl

Amazon/Zalando: bot redirect marketplace return channel. Do not Loop deep link.

Late return intl

Outside 30-day window: handoff #364 exception policy extended shipping delay.

Brexit Northern Ireland

If NI volume: dedicated RAG block or ops handoff. Bot quotes "NI policy pending validation" if absent from corpus.

Intl bot anti-patterns

Inventing domestic refund processing time for UK customer, sending incorrect zone PDF, ignoring customs_hold handoff SLA, outdated corpus FX post-Managed Markets change.

How does Qstomy deploy INT-RET-BOT on Shopify?

Qstomy deploys INT-RET-BOT: market live eligibility, deep link Loop intl, zone RAG corpus, and handoff customs hold to agents #379.

Intl return bot capabilities

  • return_intl_market_detect: ship vs return country Shopify order

  • return_intl_rag_zones: 8 INT-RAG blocks indexed EU/UK/US/CA/APAC

  • return_intl_portal_deep_link: Loop intl market-aware URL

  • return_intl_customs_pdf: downloadable CN22/23 checklist

  • return_intl_tracking_api: carrier status if RMA active

  • return_intl_handoff_379: pre-filled INT-RET ticket customs hold

Encrypted DTC scenario

EU fashion brand, 42 return_intl tickets/month US+UK, return_intl_bot_resolution 0% before Qstomy.

After INT-RET-BOT MVP: return_intl_bot_resolution 64%, return_intl_ticket_deflection -39%, customs_doc_download_bot 89/month, return_intl_handoff_rate 18%.

Explore AI support, Shopify, request a demo.

Complement #379

#380 = self-service bot country rules and documents. #379 = agent investigation customs hold and refund exception.

What is the checklist for launching the international returns bot?

Checklist INT-RET-BOT (10 steps)

  1. Audit return_intl tickets 90 days and prioritize MVP Phase 1 intents

  2. Write 8 INT-RAG corpus blocks per zone

  3. Publish /international-returns + PDF customs source RAG

  4. Connect Loop intl deep link + Shopify order context

  5. Deploy intents eligibility, label, customs_doc, delay

  6. Templates response IB-1 to IB-8 + EN markets US/UK/CA

  7. Handoff return_intl_handoff_379 → INT-RET ticket configured

  8. Widget order status intl + /international-returns trigger

  9. Anti-hallucination tests eligibility 20 orders edge cases

  10. Dashboard return_intl_bot_resolution per weekly market

In brief

  • #380 = cross-border returns AI bot, not INT-RET agents #379

  • INT-RET-BOT: 10 intents, 8 INT-RAG blocks, IB-1 to IB-8

  • MVP Phase 1: eligibility, label, customs_doc, delay

  • Bot explains documents, does not refund before receipt

  • KPI return_intl_bot_resolution: target > 60 %

FAQ

Difference with intl returns support #379?
#380 = self-service bot for country rules and documents. #379 = helper framework for agents INT-RET and macros.

Relationship with returns bot #10?
#10 = domestic RMA. INT-RET-BOT = separate cross-border intents, handoff if intl market is detected.

Can the bot generate a customs label?
Bot sends PDF template and Loop portal deep link. Label generation = portal or agent #379.

Customs hold: bot or agent?
Bot explains missing documents. Hold > 7 days → handoff INT-RET #379 with tracking.

Refund in different currency?
Bot quotes INT-RAG-FX: same order currency. Bank variance is normal, not a brand error.

Going further

Test mystery shop return intl bot: US ship order, UK customer chat « return label », check Loop deep link + PDF customs + delay 20-35 days without INT-RET agent ticket.

Share this guide #380 with product and ops: a connected INT-RET-BOT transforms opaque cross-border returns into measurable self-service by Shopify market.

Enzo

August 17, 2026

Convert over 2,000 customers on average per month with Qstomy.

The world’s 1st Shopify AI dedicated to customer conversion

Empowering 200+ e-commerce merchants

Subscribe to the newsletter and get a personalized e-book!

No-code solution, no technical knowledge required. AI trained on your e-shop and non-intrusive.

*Unsubscribe at any time. We do not send spam.

Subscribe to the newsletter and get a personalized e-book!

No-code solution, no technical knowledge required. AI trained on your e-shop and non-intrusive.

*Unsubscribe at any time. We do not send spam.