Most sanctions screening APIs look identical on the landing page: "global lists", "AI matching", "real-time". The differences that matter show up in the docs and the first week of use. Here is the checklist we would use as a buyer, with our own answers so you can compare.
1. Which lists, and from where
Ask for the publisher and the file. "Global coverage" often means a reseller's aggregate with unknown lag. KYCWire loads five files from four publishers (coverage and licences) and shows the publication date of each in the response. We do not include PEP or adverse media; if you need them, you need a second vendor.
2. Refresh cadence and failure behaviour
How often, and what happens when the publisher's site is down or the file format changes? Ours: every 6 hours; a failed load keeps the previous version active and alerts us; the status page shows staleness publicly.
3. Matching across scripts
Test it yourself: send «Ким Чен Ын» and see whether Jong Un KIM comes back. Trigram-only engines return nothing. Ask for a published test set with results — not a claim.
4. What the response explains
A score alone forces your reviewer to redo the work. Look for per-signal explanations (name similarity, identifier hit, conflicting birth year) and the matched alias.
5. Audit trail
Every check should be retrievable later with the list versions it used. Export format matters for auditors: CSV or nothing. Retention should be a plan feature you control.
6. Pricing model
Per-check pricing with a free tier lets you evaluate on real data before anyone signs anything. Enterprise-only pricing means a sales cycle before you know whether the matching is any good. KYCWire: 200 free checks a month, then $19–$149, card or crypto, cancel from the dashboard.
7. Integration surface
REST with OpenAPI, SDKs for your stack, webhooks with signatures, batch for migrations. Ours: OpenAPI 3.1, Python and Node SDKs, HMAC-signed webhooks with retries, batch up to 10,000.
Where KYCWire is the wrong choice
If your obligation includes PEP screening, adverse media, or ownership analysis (the OFAC 50 % rule), you need a broader product. If you need on-premise deployment, you need a different one. We say this on the coverage page so you do not find out after integrating.
