Bid Response (Bid Resp)

A Bid Response is the OpenRTB object a bidder returns to an exchange. It wraps one or more seatbid objects, each containing actual bids. The response can also be empty (a no-bid), but how exchanges log empty responses varies — some count them as responses, others ignore them.

Key properties of a Bid Response:

• seatbid array — holds bids grouped by buyer seat. Each seatbid has a bid array with price, creative ID, and targeting details.
• cur — currency of the bid prices (typically USD).
• nbr — no-bid reason code (optional; not all exchanges populate it).

So what?

The Bid Response is the moment a demand partner reveals intent. A high response rate means the bidder is engaged and has data to act on. A low rate means something upstream is broken — slow parsing, missing user ID, or a floor that chokes all bids.

The bid response rate is a ratio, but the denominator depends on your reporting system.

Bid Response Rate = Valid Bid Responses / Bid Requests Sent

Where:

• Valid Bid Response = a response that arrives within the timeout window and parses successfully (valid JSON, correct seatbid structure).
• Bid Requests Sent = requests the exchange dispatched to the bidder (not requests received from publishers).

Important caveats

1. No-bid counting — Some platforms count a response with an empty seatbid array as a valid response. Others treat it as a non-response. Always check your vendor’s definition.
2. Timeout window — A response that arrives after the exchange’s timeout is silently dropped and never counted as a valid response. The bidder may think it responded; the exchange disagrees.
3. Seatbid array size — A single response can contain multiple seatbids. The rate counts the response, not the number of bids inside it.
4. Error responses — HTTP 500 or malformed JSON are not valid responses, even if the bidder intended to bid.

A bid response rate of 95%+ is generally healthy for a well-tuned bidder. Below 80% warrants investigation — but don’t panic at a single low number.

What to pair it with:

• Bid Request volume — A low response rate on a tiny request pool is noise. On millions of requests, it’s a signal.
• Average response time — If response time creeps above 80% of the timeout, the rate will drop. Slow infrastructure is the #1 cause of low response rates.
• Win Rate — A high response rate with a low win rate means you’re bidding often but losing. A low response rate with a high win rate means you’re picky but effective.

Easy mistake

Someone in Slack says “Our response rate dropped to 60% — we must be losing demand!” Check timeout first. A 50ms increase in P99 latency can crater response rate without any change in demand quality.

Three levers move bid response rate: infrastructure, data quality, and auction configuration.

Infrastructure

• Server latency — The biggest lever. Shaving 10ms off P99 response time can lift rate by 5-10 points.
• Timeout alignment — If the exchange timeout is 100ms and your P95 is 95ms, you’re living dangerously. Raise timeout or lower latency.
• Connection pooling — Reusing HTTP connections reduces handshake overhead.

Data quality

• User ID match rate — If the bidder can’t resolve the user (no cookie, no device ID), it often skips responding. Improving ID sync raises response rate.
• Inventory filtering — Overly aggressive pre-bid filtering (e.g., blocking all but 5% of domains) reduces response rate by design. That’s fine — just don’t confuse it with a technical problem.

Auction configuration

• Floor price — Very high floors can cause the bidder to skip responding entirely if it knows it can’t compete. This is a strategic choice, not a bug.
• Deal priority — If the bidder only responds to deals, the open-auction response rate will be low by design.

Tradeoffs

Raising response rate at all costs is a trap. If you relax timeout limits, you may increase response rate but miss the auction deadline — those late responses are worthless. If you lower pre-bid filters, you may respond to garbage inventory and waste compute. Optimize response rate only after you’ve confirmed that the responses are competitive and arrive on time.

Bid Response rate is a platform health signal, not a demand quality metric. Treat it as a diagnostic for latency, data integration, and auction configuration.

• Always pair with average response time and win rate.
• Verify how your platform counts no-bid responses before comparing across systems.
• Optimize for consistent, on-time responses — not for a perfect 100% rate.