The Ad Waterfall
An ad waterfall is a legacy monetization method in which a publisher offers each impression to demand sources one at a time, in a fixed priority order, passing it down the chain until a partner fills it.
Key takeaways
- The waterfall calls demand sources sequentially by priority or historical price.
- Each unfilled impression 'passes back' to the next partner, adding latency.
- It systematically underpriced inventory because order, not price, decided who bid first.
- Header bidding and unified auctions largely replaced the waterfall after 2015.
How a waterfall runs
The publisher ranks partners "” direct-sold campaigns first, then networks and exchanges by expected CPM. The ad server offers the impression to the top partner; if it declines or fails to fill, a passback tag hands it to the next, and so on. The first partner willing to buy wins, regardless of whether a later partner would have paid more.
Why it lost favor
The waterfall's core flaw is that priority is a proxy for price, and a poor one. Latency compounded with every passback, and buyers deep in the chain never saw high-value impressions. Header bidding removed the sequence entirely, letting demand compete in parallel.
| Logic | Sequential, by priority / historical CPM |
|---|---|
| Failure mode | Passbacks add latency; price is not maximized |
| Successor | Header bidding / unified auction |
Frequently asked questions
Why did publishers abandon the waterfall?
Because ranking demand by priority rather than live price left revenue uncaptured and added latency. Header bidding let all demand bid at once, raising yield.
Is the waterfall still used?
Rarely for web display, but sequential mediation logic still appears in some in-app and older environments, often blended with unified-auction mechanics.