What is QPS?
QPS (queries per second) is the rate of bid requests an exchange sends to a bidder, and the main capacity limit in an RTB integration. Bidders agree a QPS cap with each exchange based on what their infrastructure and budgets can absorb, and the exchange throttles or filters traffic to stay under it. OpenRTB does not define QPS; it is an operational term.
Also called: queries per second, requests per second, bid request volume, QPS limit.
How it works
An exchange usually sees far more impressions than any one bidder can process, so it chooses which requests each bidder receives. Pretargeting rules (by country, format, device, or publisher) and the bidder's QPS cap decide that sample. Many exchanges also shift traffic based on how often a bidder bids or times out, sending more of what it buys and less of what it ignores.
QPS drives cost on both sides. Every request has to be serialized, sent, parsed, and answered within BidRequest.tmax, so the spec's advice to keep HTTP connections alive and to gzip payloads matters most at high rates. A bidder close to its limit should return a fast no-bid rather than queue requests and miss the deadline.
The common mistake is treating QPS as free. Raising the cap without the capacity to answer in time pushes up the timeout rate, and a bidder that receives traffic it never bids on lowers its own bid rate, which can lead an exchange to cut its share. Measure bid rate and timeout rate per exchange before asking for more QPS.
Read more
Related terms
Every term is in the glossary.