Chargeback disputes where both sides have data and neither has evidence

Vendor-compliance disputes have a signature failure mode: both sides are drowning in data and neither is holding evidence. Your system logged a receipt. The supplier's system logged a shipment. Each record is authoritative exactly as far as the wall of the system that produced it — and the dispute happens in the space between the walls. So it stalls, and the deduction stands or falls on leverage rather than facts.

The fix is not more data. It is a record either side can verify without joining the other's anything — at which point the standoff becomes a lookup. Here is the same dispute, replayed both ways.

The standoff anatomy

A model short-ship claim, in its natural habitat. PO 7719, one supplier, one DC. The supplier invoices 96 cases. Your DC's receiving system shows 91. Compliance deducts for 5 cases plus the handling fee, the supplier disputes, and the file lands on both companies' analysts.

Replay one: exports at dawn

The supplier produces a CSV from their WMS: 96 cases staged, 96 loaded, seal number, timestamp. Your team produces a CSV from the receiving system: 91 cases scanned at the door, timestamp, lane. Both extracts are internally consistent. Neither is checkable by the other side — each is a report about a system the counterparty cannot inspect, prepared by the party it favors.

So the analysts do what the structure forces: they argue provenance. Was the supplier's load count a scan or a pick-list assumption? Did your door scan miss cases that arrived? Was a pallet diverted at cross-dock? Six emails become a call, the call becomes a standing agenda item, and ninety days later the dispute closes on a split — not because anyone learned what happened, but because both sides priced the argument above the 5 cases. Multiply by every deduction line, and this is a department.

Replay two: computed identity plus attested observer

Same claim, different record. Both the shipping event and the receiving event are conformant EPCIS 2.0 events whose identities are their CBV 2.0 §8.9 hashes — computed from content by a published algorithm, so either party recomputes the digest and confirms the record is what it was when it was captured. Tamper with a quantity after the fact and the identity breaks loudly; the computed-identity brief shows the one-character demonstration.

And the receiving event answers the question replay one died on, because it carries a performer. EPCIS 2.0 §7.2.2 defines the standard's five event dimensions — what, when, where, why, how — and no performer is among them; its party fields are organisation-grain (CBV 2.0 §8.7.1). Ten minutes, the specs are public, don't take our word for it. Our record completes the missing grain twice over: who, the attested observer of the receipt — the dock associate's verified identity, or the agent working the lane — and capturedBy, the warrantor account that stands behind the capture. Two grains, never collapsed, because an observer can scan under an account it does not own, and a dispute needs both answers.

Now the exchange runs like this. Supplier: here is the shipping event, 96 cases, hash attached — verify it. You: here is the receiving event, 91 cases, hash attached, attested observer attached — verify it. Both records survive verification, so the facts are agreed in minutes: 96 left, 91 arrived, and the investigation collapses to the one real question — what happened in between — with the carrier's handoff as the obvious next record to pull. No provenance argument, because provenance is computed. The dispute is smaller, faster, and about the world instead of about the databases.

The asymmetry

Here is the operational point for a compliance director: whoever holds verifiable evidence sets the dispute's tempo. When your receiving record verifies independently — without the supplier joining anything — your deduction letters change character. They stop being claims and start being exhibits. Suppliers contest fewer of them, and the contests that remain are the ones with a genuine question inside, which is what your analysts should be spending their days on.

Require it, and offer it

The end state is both sides holding evidence, because a dispute between one evidence-holder and one export-holder still costs you analyst time. So the program move is symmetrical: require hash-identified, performer-attributed event records from suppliers on your dispute-heavy lanes — and offer them the same in return, as a shareable dispute-evidence view scoped to the transaction in question. A supplier who can verify your record has far less to argue with; a supplier who can produce their own has far less to fear from you. Adversarial process, cooperative substrate.

Your dispute ledger already knows which lanes to start with. Get started — the retail branch of the interview asks about the last claim you filed and the evidence you attached, how you decide today whether a loss is theft, damage, or a scan error, and how many of these you see a month. It takes minutes.