Practical guide

Sanctions screening API cost: model the full workload

Estimate screening volume, ongoing monitoring, review effort, and integration costs without confusing search traffic, API calls, and completed subjects.

On this page

Start with the billing unit

An HTTP request, a submitted subject, a completed screening, and a human case review are different units. Ask what the provider charges for before comparing plan prices. A batch can contain many subjects, while reopening retained evidence may not create another screen.

SanctionsKit’s published product content describes a shared production subject allowance across interactive screening, batches, and monitoring. Check the current pricing page and usage documentation for commercial terms rather than relying on an old quoted price.

Separate one-time and recurring demand

Model onboarding checks, portfolio reviews, scheduled monitoring, source-triggered rescreens, and corrected-subject operations. Include a realistic margin for growth and policy changes. Do not count an unchanged idempotent replay as a new business subject.

Scheduled monitoring can dominate the workload. Ten thousand retained customers on a daily schedule create roughly 300,000 scheduled rescreens in a 30-day month, before source-triggered activity. That arithmetic describes an illustrative workload, not a price quote or a promise about exact run counts.

Build a transparent scenario

Synthetic planning example: 2,000 new subjects, 500 ad hoc checks, and 1,000 monitored subjects checked daily in a 30-day period yield 32,500 planned screens before source-triggered runs and other additions. A six-hour schedule changes that monitor component substantially.

Keep source-triggered demand as an explicit variable instead of silently assuming it is zero. Track actual usage by workflow after launch and refine the forecast.

planned_subjects = onboarding + ad_hoc + portfolio_checks
                 + monitored_subjects * planned_runs_per_subject
                 + source_triggered_rescreens
                 + corrected_subject_operations

Example: 2,000 + 500 + 0 + (1,000 * 30) = 32,500
Add source-triggered and other separately planned activity.

Include the cost of a usable review process

A cheaper comparison endpoint may leave the team to build assignment, candidate review, independent approval, evidence retrieval, and audit exports. Include those engineering and operational costs in a build-versus-buy comparison.

Review effort also depends on data quality and policy, not simply the number of alerts. Measure how many cases require additional evidence and how long they remain unresolved. Lowering sensitivity to reduce analyst work can hide missed candidates unless the change is tested against appropriate cases.

Plan for caps and failure handling

Distinguish a temporary rate limit from a plan usage cap, and document the action for each. In SanctionsKit, monitoring shares the allowance with interactive screens and no automatic overages are charged under the published model. A quota issue should not quietly clear a customer or erase a monitoring failure.

Use idempotency for accepted-request retries and isolate failed batch rows. Preserve enough operation state to avoid duplicate work after a timeout. Retention and monitoring obligations also continue to need explicit handling when a relationship ends.

Compare providers on a consistent worksheet

Evaluate the actual required sources and supported subject types before comparing per-unit costs. A lower price for less coverage is a different product, not automatically a saving.

  • Required coverage, freshness evidence, and unavailable-source behavior.
  • Completed-subject billing, retries, batches, and monitoring consumption.
  • Included reviewer seats, approvals, retained evidence, and exports.
  • Retention options, expiry behavior, and monitor lifecycle.
  • Rate limits, usage caps, support boundaries, and integration ownership.

Official references