One API call vs three providers — the math for a credit pipeline

The three-provider stack

A standard build-from-scratch credit-decision pipeline integrates three categories of verification provider:

1. SOS data. A provider that exposes Secretary-of-State data across multiple states, either through a unified API or state by state.
2. FMCSA / USDOT data. A provider that exposes SAFER and operating-authority data, typically by reselling the FMCSA Census or by pulling SAFER directly.
3. OFAC / sanctions screening. A provider that screens names against the OFAC Consolidated List and often other sanctions and PEP databases.

Each provider has its own pricing, its own SLA, its own JSON schema, its own error handling, and its own rate limits. Stitching them together is the engineering exercise that an SDK or in-house wrapper typically resolves.

The question for a developer building a verification pipeline: is the three-provider stack cheaper, faster, or better-supported than a single bundled call?

The latency math

Three separate API calls take longer than one bundled call by approximately the latency of two of them — because the bundled call runs the upstream queries in parallel server-side.

A rough latency profile for a typical verification:

• SOS query: 5-15 seconds for most states, 15-30 seconds for the heavier-portal states like California and Nevada.
• FMCSA query: 1-3 seconds. FMCSA’s SAFER endpoints are fast.
• OFAC query: 50-500 milliseconds. OFAC matches against a local copy of the Consolidated List; it’s a lookup, not a network call.

Three serial calls: 6-18 seconds for fast SOS, 16-34 seconds for slow SOS. Three parallel calls: max(SOS, FMCSA, OFAC) — dominated entirely by SOS latency.

For the slow-state case, parallel is roughly half the total time. The difference is meaningful if your pipeline is user-facing (a credit-check button in an underwriting UI) and unimportant if it’s batch (an overnight cron).

The cost math

Per-query pricing varies significantly. Rough 2026 market rates for KYB providers:

• SOS data: $0.50-$5.00 per state per query, depending on volume and state-difficulty. Slow states (California, Nevada) cost more because they are simply harder to reach.
• FMCSA data: $0.05-$0.30 per query. Cheap because SAFER is fast and free upstream.
• OFAC data: $0.02-$0.20 per query. Cheap because the data is downloadable and refresh is daily.

Three-provider stack per query: $0.57-$5.50 depending on state.

Bundled providers typically price at a flat rate per verification, often $0.50-$3.00 depending on state, with a markup that’s usually below the sum of the three component costs because the bundle provider amortizes their infrastructure costs across all calls.

For most volumes, the bundled cost is lower than the sum of the three components — sometimes meaningfully lower, sometimes only marginally. At very high volume (10K+ queries/day), the spread narrows because per-component providers offer volume discounts that bundles may not match.

The engineering math

Here is where the stack actually breaks even or doesn’t. Three providers means:

• Three SDKs to maintain. Or three sets of REST-API wrappers to keep current. When a provider deprecates an endpoint, you have to handle the migration.
• Three error-handling regimes. Provider A returns 429 for rate limiting; provider B returns 503; provider C returns 200 with {"error": "rate_limited"} in the body. Each needs to be handled differently.
• Three retry strategies. Different providers have different idempotency guarantees, different backoff requirements, and different per-status retry behavior.
• Three logging surfaces. Tracking latency, success rate, and cost across three providers requires either three separate observability integrations or a unified wrapper that normalizes them.
• Three SLAs to track. If one provider has a bad day, your pipeline has a bad day — but the failure mode is unclear to the operator unless you’ve instrumented per-provider.

Engineering cost is hard to amortize against per-query pricing. A senior engineer spending one week per quarter maintaining provider integrations is $40K-$80K per year in fully-loaded cost. That’s 13-26 million queries’ worth of margin difference at a $0.003-$0.005 per-query spread. For most companies, the engineering tax exceeds the per-query savings of best-of-breed providers.

When the three-provider stack does make sense

A few cases where the build-from-three approach is the right call:

You’re a sanctions-compliance specialist. If OFAC screening is your core business, you probably want OFAC depth (PEP lists, adverse-media screening, watchlist coverage) beyond what a bundled provider offers. Direct OFAC-provider integration is the right call for that specific capability.

You have specific data residency requirements. Some providers offer EU-only or US-only data residency that a bundled provider may not. Compliance constraints can override the engineering-cost argument.

You have very high volume and very specific state mix. At scale and with a known geographic distribution, you can negotiate volume discounts at the component level that don’t apply to the bundle. A company doing 1M California-only queries per month can probably get California cheaper from a specialized California provider than from a national bundle.

You’re building infrastructure you intend to license to others. If your verification pipeline is your product, you may want to own the upstream integrations to differentiate.

For most credit-decision pipelines, none of those conditions apply. The bundled call is operationally simpler, often cheaper, and almost always faster.

What VerifySOS exposes

The /api/v1/lookup endpoint accepts a company name and state code and returns a single JSON response with:

• The SOS record (entity name, status, agent, principal address, officers as filed).
• The FMCSA record when the entity has a USDOT, including operating-authority status, MCS-150 date, BASIC scores.
• The OFAC screen result against the Consolidated List, with name matches and designation specifics.
• Cross-state officer search results (when officers appear on other SOS records).
• Any flags surfaced by the verification logic (stale-date warnings, standing transitions, address overlap, nominee patterns).

One call, one auth header, one JSON schema. No SDK to maintain. Rate limit is enforced at the API gateway. Latency dominated by upstream SOS performance.

What this means for you

If you’re building a verification pipeline from scratch, do the math on engineering cost, not just per-query cost. The provider mix that’s cheapest per query is rarely the cheapest end-to-end once you account for ongoing integration work. The bundled call is almost always the better-fit default, with build-from-components reserved for specific cases where you actually need the depth that only specialized providers offer.

A free lookup shows the bundled output. The same payload is at /api/v1/lookup for developers.