Skip to content

CONCEPT Cited by 1 source

Evidence library for disputes

Definition

An evidence library for disputes is a merchant-curated document corpus — terms and conditions, return policies, service agreements, and similar boilerplate evidence — uploaded to the dispute-management system once and auto-selected per dispute based on the dispute's characteristics.

Disclosed by systems/stripe-smart-disputes at the 2026-05-27 Stripe Sessions roundup. Per the post:

"With the evidence library, you upload and store these documents once, and Smart Disputes automatically selects and includes them in your evidence packet based on the dispute's reason code, network requirements, and cardholder claims — no manual resubmission needed."

Auto-selection keys

Three orthogonal selection keys are named:

  1. Reason code — the card network's classification of the dispute (e.g. 4853 — Cardholder dispute of merchandise/ services, 4855 — Goods or services not provided).
  2. Network requirements — Visa / Mastercard / Amex / etc. each have their own evidence-format and required-document rules.
  3. Cardholder claims — the specific complaint the cardholder filed (non-receipt, not as described, unauthorized, etc.).

Selection is the Cartesian product — for each (reason code, network, claim) tuple, Smart Disputes selects the appropriate subset of stored documents and assembles the evidence packet.

Architectural shape

The architectural shape is evidence-library with auto-selection: per-document-type semantic-routing automation over a static merchant-supplied corpus, gated by multiple orthogonal keys.

This generalises beyond disputes to any submission-bundle workflow with composable per-case requirements:

  • Compliance-document submission (different jurisdictions require different documents).
  • Insurance-claim filing (different claim types require different supporting documents).
  • KYC / KYB verification (different account types, different countries).

Why this matters

Pre-evidence-library, every dispute required manual evidence assembly:

  • Operator looks up the dispute reason code.
  • Looks up the network's evidence-format requirements.
  • Pulls the right documents from a Drive / SharePoint folder.
  • Bundles into a packet matching the network's submission format.
  • Submits within the network's response window.

The work was identical across thousands of disputes because the underlying documents (T&C, return policy, service agreement) don't change per dispute — only which subset is needed and how it's packaged. Auto-selection collapses the manual work to zero for documents already in the library.

Combined with systems/stripe-smart-disputes's AI-recommended-evidence feature, the merchant's role narrows to (1) keeping the evidence library fresh and (2) supplying case-specific evidence (tracking numbers, customer usage logs) that the recommended-evidence system surfaces.

Caveats

  • Selection mechanism not disclosed. Rule-based? Learned? LLM-classified? The post says "automatically selects" without specifying.
  • Versioning / freshness policy not described. When the merchant's terms change, how is the library updated? Are old document versions preserved for disputes from the period they were active?
  • Format-conversion handling not described. Does the library auto-format documents to network-required formats?
  • Coverage envelope undisclosed — which document types are supported beyond the named three (T&C, return policies, service agreements)?

Seen in

Last updated · 542 distilled / 1,571 read