Bids and bidder activity
Bid responses, wins, no-bids, and timeouts, with bidder and ad-unit identifiers.
- Bid value & currency
- Response time
- Size & media type
Point the Prebid script you already load at your BidTune URL. Your bidders, ad server, and partners stay in place. BidTune serves your existing bundle and runs measured experiments on a slice of traffic.
One script. One URL to change.
When your site is activated, set the src below. BidTune serves your existing bundle from that URL. It is not a second Prebid on the page.
<script src="https://cdn.example.com/prebid.js"></script><script src="https://s.bidtune.net/YOUR_SITE_ID/p.js"></script>Example URLs. Use the installation details supplied for your site, including any integrity or security-policy requirements.
No ad-server access is required. Your bidders, ad server, and partners stay in place. A managed setup still needs permission to change the script URL and the settings in scope.
First findings land in about 2 days. Each test needs at least 72 hours and 2,000 sessions in each arm. A site-wide test needs about 7,408 sessions over 14 days (about 530 a day). The smaller arm is 30% of the traffic outside the 10% holdback.
Sign in and add your site. BidTune gives you one script tag to install.
Use your BidTune URL in place of your current Prebid script. Your stack stays in place.
Confirm traffic is arriving. BidTune tests one change at a time and records the measured lift.
Bidders, demand relationships, and the ad server stay in place.
Every change BidTune made, with its measured lift.
Set the scope, pause the work, or undo a change.
BidTune tries one change, measures it against your current configuration, and keeps it only when it is measurably better.
That slice keeps your current configuration, untouched. It is the comparison for the whole setup.
A test runs there. It needs at least 72 hours and 2,000 sessions in each arm.
Every change BidTune made, with its measured lift. A change that is not measurably better rolls back.
BidTune listens to supported Prebid events and connects them with delivery and configuration context.
Bid responses, wins, no-bids, and timeouts, with bidder and ad-unit identifiers.
Auction duration, requested bidders, and the effective timeout when available from Prebid.
Non-empty GPT render events and available targeting, bidder, and creative identifiers.
Connect events with page path, device, country, visit, configuration, and experiment assignment.
A winning bid is not the same as a rendered ad. Available events and fields depend on the integration. BidTune uses observed renders and available price signals to evaluate delivery. Read the measurement approach.
A site-scoped identifier in browser session storage links activity within a visit. Review data collection alongside your existing consent setup before enabling the integration.
An MCP server is planned, so you can read BidTune’s bid data from the AI tools you already use and bring a question back. That connector is not available yet.
Until then, experiments run inside BidTune: 30% of optimized traffic, a 10% holdback, and the ledger.
MCP connector · Planned, not available“Would a longer timeout capture late bids without slowing the page too much?”
Example question. The connector that would accept it is planned, not available.
After the script is collecting, BidTune reads the auction and can test one change on 30% of optimized traffic. The 10% holdback stays on your current configuration. Kept changes show up in the ledger with their measured lift.
Your 14-day trial begins when your site is activated and collecting data.
14 days free after activation. No credit card to get started.