Caching and rate limits

Browsers and CDNs can cache successful post and rich snippet responses, and all Storefront endpoints share one rate limit per publishable key.

Successful post and rich snippet responses use these cache durations:

• Browsers can cache a response for 60 seconds.
• Shared CDNs can cache a response for five minutes.
• CDNs can serve a stale response for five more minutes while refreshing it.

The complete request URL and publishable key identify the cached response. Rich snippets also vary by referrer because the page URL can select a product or market.

Error responses are not publicly cached. Content removed in Cevoid can remain in a shared cache for up to ten minutes.

Review responses, including ratings and summaries, use Cache-Control: private, no-store, so browsers and CDNs never cache them.

All Storefront endpoints share one limit per publishable key: 20 requests per second, with bursts of up to 40 requests. Each request counts once, regardless of page size.

For example, after a burst of 40 requests uses the available allowance, one second without requests restores room for 20 more. Two seconds restores the full allowance of 40.

When the allowance runs out, the API returns 429. Wait the number of seconds in Retry-After before retrying. RateLimit-Limit is the burst allowance (40), RateLimit-Remaining is the available allowance, and RateLimit-Reset is the retry delay (0 on successful requests).

Only requests that reach the API count. Responses served directly from a CDN cache do not.