Picture a single ad slot worth real money. Before header bidding, a publisher filled it the way a hand-me-down gets passed down a family: offer it to the first partner, and if they pass, offer it to the next, and the next, until someone takes it.

That setup is the waterfall (also called a daisy chain). It is the old way, and header bidding was invented specifically to beat it. So the goal here is simple: understand exactly why the waterfall leaves money behind.

How a waterfall works

Demand sources are stacked in a fixed order. The ad server offers the impression to source 1. If source 1 passes (no fill, or a bid below the floor), the tag passes back and the impression goes to source 2, then source 3, and so on until something fills. One source at a time, in order.

Why it under-monetizes

• Priority, not price. Tiers are ordered by a partner’s historical average CPM or a business relationship, not by what each source will actually pay for this specific impression. A source two tiers down might value this exact user far more.
• First acceptable bid, not highest bid. The first tier that fills wins, even when a lower tier would have paid more. There is no head-to-head competition.
• Latency cascade. Each passback is another network round-trip. Deep waterfalls add seconds of delay, and users abandon slow pages, so impressions are lost outright.
• Opacity. The publisher cannot see what the tiers below the winner would have bid, so it has no signal to reorder or reprice intelligently.

Worked example

A four-tier waterfall, ordered by historical average CPM. Each tier’s fill rate is the chance it takes the impression when asked:

Because order wins, Network A clears most impressions at $2.00. Network B, which would have paid $3.50, only ever sees the 40% that A passed back. The waterfall systematically leaves the higher bid on the table whenever a lower-paying source sits higher in the stack.