API Changelog

The reliability and passRate7d fields in GET /v1/states now measure source availability: the percentage of lookups where the state portal was reachable. Previously, these fields counted only billed lookups as successes, which meant a lookup that correctly determined an entity does not exist counted as a failure. A lookup that reaches the portal and returns not-found is now counted as a success. Only source_unavailable (portal unreachable or timed out) counts as a failure. This change more accurately reflects scraper infrastructure health.

Every lookup response now includes a searchedName field containing your original search input. Compare it against data.entityName to verify the returned entity matches your query. This is an additive field and does not change existing response fields.

Cached data lookups now use exact name matching instead of substring matching. Previously, a search could return a similarly-named but different entity (for example, searching "Airbnb Inc" could return "Nos Center Airbnb Inc." from cache). This no longer happens. Some lookups that previously returned a cached match will now return no match and fall through to a live lookup. This is the correct behavior: the cache only returns your exact entity, not a partial name match.

The default maximum age for cached data was reduced from 60 days to 30 days. Cached results older than 30 days now fall through to a live lookup instead of being served from cache. This ensures fresher data for cache-first and cache-only requests. If you have a custom maxCacheAgeMonths setting, it is unchanged.

Lookups in Texas and Illinois that return a verified not-found result (the entity does not exist in the state registry) are now billed at your standard lookup rate. Previously these were not charged. This applies only when the state portal confirms the entity does not exist, not when the portal is unavailable or the lookup times out. Failed lookups are never charged.

Lookups with cache_preference: "cache_only" for California now return cached data even when the live scraper circuit breaker is open. Previously, a cache-only request would return SOURCE_UNAVAILABLE during scraper outages despite cached data being available.