Transfon BiddingStack Launches Placement Level A/B Testing for Publishers
by on 10th Aug 2026 in News
BiddingStack by Transfon, the unified header bidding and ad management platform, today announced the general availability of placement level A/B testing. Publishers can split live traffic into stable groups, compare refresh settings, ad sizes, lazy-load thresholds, bidder line-ups and floor price strategies, then roll out the winner from the console. BiddingStack is licensed on a SaaS model.
BiddingStack is a full-stack, AI-powered ad management platform that runs Prebid.js, Amazon TAM, Google Open Bidding, and Google AdX as one unified stack:
- Multi-format bidding across banner, native and video
- Web and mobile apps, the latter through native mobile SDKs
- A single console to tune yield across all of it
- For publishers bringing header bidding in house, an alternative to other prebid wrappers
Changes get made, but rarely proved
Most ad stack decisions are settled by argument. Someone shortens a refresh interval, adds a bidder or raises a floor, and next week's revenue becomes the verdict. Seasonality and traffic mix move that number too, so nobody can say which change did what. On a managed service there is a second problem. The partner recommending the change also writes the report on whether it worked.
Stable groups, tested on live traffic
- Testing groups put visitors into stable, non-overlapping buckets, and the split holds for the life of the test
- Tests run in parallel without interfering: one placement can trial a floor price strategy while another tries a different bidder line-up
- Publishers read the outcome in their own reports, measured on revenue, RPM and viewability
Results read at bidder level, in realtime
Any group breaks out by demand partner through BiddingStack's realtime bidder level insight:
- Bid rate, win rate, timeout rate, CPM and revenue, for every bidder, client side, and server side
- Figures move as traffic arrives
Aggregates hide things. A variant might lift RPM while quietly pushing one partner into timeouts, or win more impressions at a lower CPM. Both show up next to the headline number.
Anything configurable is testable
Tests are set at placement level and cover the settings that move yield:
- Refresh and viewability rules
- Ad sizes and insertion types
- Lazyload and preload thresholds
- Bidder line-ups
- Dynamic floor price strategies
Every configuration change is versioned and diffable, so a result traces back to the setting that produced it, and a losing variant reverses cleanly.
Built for publishers running their own stack
A/B testing joins the controls BiddingStack already puts in the console instead of behind a support ticket:
- Refresh and viewability management
- Dynamic floor pricing
- Bidder traffic optimisation
- Delivery rules
Every partner in the auction is the publisher's own, on contracts they hold and terms they negotiated.
An independent wrapper, on a SaaS model
BiddingStack is an independent wrapper on a SaaS model, with a flat fee and no revenue share. Every point of RPM a test wins belongs to the publisher. Independence counts for the testing too, because the platform has no stake in which bidder takes the impression.
"Most of what passes for optimisation in this industry is untested opinion, and the vendor offering it usually marks its own homework," said Bruce Dou, CEO and founder of Transfon. "Give a publisher the tools to split traffic, run the test on their inventory and read the result in their reports, and the argument is over. It settles on evidence. That only counts coming from an independent wrapper, which is why we work on a SaaS model."
Availability
Placement level A/B testing is generally available today. Existing customers can create testing groups from the dashboard. Publishers on another wrapper, or weighing a migration, can get in touch at www.biddingstack.com for a no-obligation audit and a trial on part of their inventory.
Transfon
BiddingHeader BiddingPrebidSaaS
Follow ExchangeWire