Return-to-Refund: Object-Centric Process Mining Analysis

1. 0Executive summary
2. 1Method and assumptions
3. 2Event log and structure
4. 3Process model and variants
5. 4Cycle time and conformance
6. 5Credit issued, never refunded
7. 6Exception registers
8. 7Data quality
9. 8Findings and actions
10. ASuiteQL used
11. BHand-check

0Executive summary

Returns are new to this account — the first return authorization is dated 2026-08-05 — and all four RMAs belong to one customer, Design Excellence Ltd., each returning a partial quantity from a seven- or five-unit sales order at $51.52 a unit. Small as it is, the population exercises the whole reference path once and breaks it three ways:

1. The one complete chain closes in zero days. RMA23: SO4266 (2026-08-01) → RMA23 → IR1175 → CM06 → Refund 1, all on 2026-08-05. Authorisation, receipt, credit and refund on the same day, $154.56 to the cent. That is the standard the other three should be measured against.
2. Goods received, credit never issued. RMA24 (2026-08-16, $103.04): the return was received the same day (IR1178) and has sat in Pending Refund for 19 days with no credit memo. Two further credit memos with no RMA — CM01 $108.48 (Susan Adams) and CM03 $152.43 (Bobby Davis) — are open and unapplied. Total customer money owed with nothing moving: $363.95.
3. Multi-object structure is real but tiny. SO4268 (7 units) has two RMAs against it — RMA25 for 3 units (pending receipt) and RMA26 for 1 unit (pending approval) — so the order perspective sees a fan-out the RMA perspective does not. All three quantity-bearing RMAs are partial returns (3 of 7, 2 of 5, 3 of 7 + 1 of 7). No RMA spans more than one order, and no RMA links to an invoice: returns here are authorised against the sales order.

Separately, the single cash-sale refund CR01 returns $716.39 against a $241.70 cash sale — a $474.69 over-refund that the link table makes visible in one query.

1Method and assumptions

2Event log and structure

3Process model and variants

Figure 1 — Directly-follows graph, all return cases. Emerald = the complete path (RMA23); grey = open prefixes and exceptions. Labels: count.

4Cycle time and conformance

5Credit issued, never refunded — the actionable cohort

6Exception registers

6-A · Open return authorisations (3)

6-B · Amount anomaly (1)

The cash sale's payment is itself unapproved. A refund of 3× the sale value against an unapproved sale is either a data-load artefact or a control gap; either way it is the single largest return-side figure in the account.

7Data quality

8Findings and recommended actions

1. Clear the $363.95 stalled cohort today. Create the credit memo for RMA24 (received 19 days ago), apply or refund CM01 and CM03. Then add a saved search — RMAs in Pending Refund with a linked receipt and no credit memo; credit memos open more than 7 days — to the AR dashboard so the stall is visible the day it happens.
2. Investigate CR01. A $716.39 refund against a $241.70 unapproved cash sale. Confirm whether the refund amount is a load error; if not, establish who authorised a refund at three times sale value.
3. Keep the RMA23 standard. Authorise, receive, credit and refund in one day is achievable — it happened. Measure every future return against it; when volume grows, the same queries in Appendix A produce the distribution.

AAppendix A — SuiteQL used

Q1 — Census with live status labels

SELECT x.type, x.status, x.docs, BUILTIN.DF(t.status) AS label, x.first_date, x.last_date, x.total
FROM (SELECT type, status, COUNT(*) AS docs, MIN(id) AS sample_id, TO_CHAR(MIN(trandate),'YYYY-MM-DD') AS first_date,
             TO_CHAR(MAX(trandate),'YYYY-MM-DD') AS last_date, ROUND(SUM(ABS(foreigntotal)),2) AS total
      FROM transaction WHERE type IN ('RtnAuth','CustCred','CustRfnd','CashRfnd') GROUP BY type, status) x
JOIN transaction t ON t.id = x.sample_id ORDER BY x.type, x.docs DESC

Q2 — Link-type census

SELECT p.type AS prev_type, n.type AS next_type, l.linktype, COUNT(*) AS link_rows, COUNT(DISTINCT l.previousdoc || '-' || l.nextdoc) AS pairs
FROM nexttransactionlinelink l JOIN transaction p ON p.id = l.previousdoc JOIN transaction n ON n.id = l.nextdoc
WHERE p.type IN ('RtnAuth','CustCred','CustRfnd','CashRfnd') OR n.type IN ('RtnAuth','CustCred','CustRfnd','CashRfnd')
GROUP BY p.type, n.type, l.linktype ORDER BY COUNT(*) DESC

Q3 — Every return document with inbound / outbound links

SELECT t.id, t.type, t.tranid, TO_CHAR(t.trandate,'YYYY-MM-DD') AS trandate, BUILTIN.DF(t.status) AS status_label, BUILTIN.DF(t.entity) AS customer,
       ROUND(ABS(t.foreigntotal),2) AS total, ROUND(t.foreignpaymentamountunused,2) AS unused, t.memo,
       (SELECT LISTAGG(p.type || ' ' || p.tranid || ' [' || l.linktype || ']', '; ') WITHIN GROUP (ORDER BY p.trandate)
          FROM nexttransactionlinelink l JOIN transaction p ON p.id = l.previousdoc WHERE l.nextdoc = t.id AND l.linktype <> 'CostRtrn') AS inbound,
       (SELECT LISTAGG(n.type || ' ' || n.tranid || ' [' || l.linktype || ']', '; ') WITHIN GROUP (ORDER BY n.trandate)
          FROM nexttransactionlinelink l JOIN transaction n ON n.id = l.nextdoc WHERE l.previousdoc = t.id AND l.linktype <> 'CostRtrn') AS outbound,
       (SELECT MAX(tl.createdfrom) FROM transactionline tl WHERE tl.transaction = t.id AND tl.mainline='T') AS createdfrom
FROM transaction t WHERE t.type IN ('RtnAuth','CustCred','CustRfnd','CashRfnd') ORDER BY t.trandate, t.id

Q4 — Quantities on RMAs, receipts and originating orders

SELECT t.type, t.tranid, TO_CHAR(t.trandate,'YYYY-MM-DD') AS trandate, BUILTIN.DF(t.status) AS status_label, ROUND(ABS(t.foreigntotal),2) AS total,
       (SELECT COUNT(*) FROM transactionline tl WHERE tl.transaction=t.id AND tl.mainline='F' AND tl.taxline='F') AS lines,
       (SELECT SUM(ABS(tl.quantity)) FROM transactionline tl WHERE tl.transaction=t.id AND tl.mainline='F' AND tl.taxline='F') AS qty,
       (SELECT SUM(NVL(tl.quantityshiprecv,0)) FROM transactionline tl WHERE tl.transaction=t.id AND tl.mainline='F' AND tl.taxline='F') AS qty_recv_or_shipped
FROM transaction t WHERE t.id IN (40200,40303,40304,42234) OR t.tranid IN ('IR1175','IR1178') OR t.type = 'RtnAuth' ORDER BY t.trandate, t.id

BAppendix B — Hand-check

RMA23 $154.56 = 3 units × $51.52 ✓ · CM06 $154.56 = RMA23 ✓ · Refund 1 $154.56 = CM06 ✓ · RMA24 $103.04 = 2 × 51.52 ✓ · RMA25 + RMA26 = 3 + 1 = 4 of 7 units on SO4268; $154.56 + $51.52 = $206.08 ✓ · Owed = 103.04 + 108.48 + 152.43 = $363.95 ✓ · CR01 excess = 716.39 − 241.70 = $474.69 ✓ · Documents: 4 + 2 + 3 + 1 + 1 = 11 return-side incl. receipts; 9 excluding receipts ✓.