What Is Waterfall in Advertising?
In programmatic advertising, a waterfall (also called daisy chaining or sequential auctions) is an inventory monetization method where a publisher offers unsold ad impressions to demand partners one at a time, in a pre-set priority order. Each network or exchange gets exclusive first access. If it passes or bids below the floor price, the impression drops to the next tier. This continues until a buyer accepts or the impression goes unfilled.
The waterfall dominated digital advertising from roughly 2009 through 2016, shaping how billions of impressions were traded daily before header bidding disrupted the model.
How the Waterfall Works
Publishers configure their ad server (historically Google Ad Manager) with a ranked list of demand sources. The impression enters at the top and cascades downward until it clears or expires.
Typical Waterfall Structure
| Tier | Demand Source | Floor Price | Outcome |
|---|---|---|---|
| 1 | Direct-sold campaigns | $15.00 CPM | No available line item — passes |
| 2 | Premium network (e.g., Criteo) | $6.00 CPM | Bids $4.20 — below floor, passes |
| 3 | Mid-tier exchange | $2.00 CPM | Bids $2.50 — wins |
| 4 | Remnant network | $0.50 CPM | Never reached |
The key mechanic is exclusivity. Unlike header bidding, where multiple buyers compete simultaneously, the waterfall gives each partner a sequential monopoly window on the impression.
Revenue Calculation in a Waterfall
Publishers estimate waterfall revenue using expected fill rates and CPMs at each tier:
Expected Revenue = (Fill Rate Tier 1 × CPM Tier 1) + ((1 – Fill Rate Tier 1) × Fill Rate Tier 2 × CPM Tier 2) + …
Example with two tiers:
- Tier 1: 40% fill rate at $5.00 CPM = $2.00 expected CPM
- Tier 2 (receives remaining 60%): 70% fill rate at $1.50 CPM = $0.63 expected CPM
- Blended expected CPM: $2.63
This calculation reveals the core problem. Demand sources are ranked by historical averages rather than live bid values. That means a Tier 3 buyer willing to pay $8.00 CPM on a given impression may never see it if Tier 1 fills at $5.00 first.
Why Publishers Used Waterfalls
The waterfall emerged from practical constraints of early ad tech. Publishers needed a manageable way to monetize unsold inventory after direct sales without overwhelming their ad servers with simultaneous demand calls. A sequential system kept latency predictable and gave large premium networks the negotiating power to secure top-tier placement.
Major publishers including Vox Media and Condé Nast ran waterfall configurations through the early 2010s, relying on networks like Tribal Fusion, ValueClick, and early iterations of Google’s AdX at descending tiers. Publishers with strong direct-sold programs found the waterfall acceptable because remnant inventory was genuinely lower value. The problem intensified as programmatic buying matured and the gap between waterfall CPMs and theoretical auction-clearing prices widened.
The Core Inefficiencies
Latency
Each sequential call adds server response time. A waterfall passing through six demand sources could add 600 to 1,200 milliseconds of page load time before serving an ad. Research from Google’s DoubleClick (now Google Ad Manager) found that a one-second delay in ad loading correlates with measurable drops in viewability and user engagement.
Revenue Leakage
Because tier rankings relied on historical average CPMs rather than real-time bids, a buyer with a high-value audience match in Tier 4 could never outbid a lower-paying Tier 1 buyer. AppNexus (now Xandr) estimated in 2016 that publishers using pure waterfall setups were leaving 20 to 40 percent of potential programmatic revenue on the table compared to simultaneous auction methods.
Opacity
Demand partners at lower tiers had no visibility into why they received certain inventory. This made optimization difficult and gave top-tier networks a built-in advantage regardless of actual bid competitiveness.
Waterfall vs. Header Bidding
| Factor | Waterfall | Header Bidding |
|---|---|---|
| Auction structure | Sequential, exclusive | Simultaneous, competitive |
| Price discovery | Historical averages | Real-time bids |
| Latency | Higher (stacked calls) | Lower (parallel calls) |
| Publisher revenue | Generally lower | Generally higher |
| Setup complexity | Low | Moderate to high |
| Common use today | Mobile in-app, CTV | Web display, video |
Where Waterfalls Still Operate
Despite the shift toward header bidding on web inventory, waterfall logic remains common in two environments:
Mobile In-App Advertising
Mobile apps often use mediation platforms such as ironSource, MAX by AppLovin, or Google AdMob to manage demand sources. These mediation layers traditionally used waterfall configurations, though in-app bidding (the mobile equivalent of header bidding) is now growing. As of 2023, AppLovin reported that publishers using MAX’s simultaneous bidding features saw average revenue increases of 20 to 30 percent over waterfall-only setups. The gap persists even in mobile.
Connected TV (CTV)
CTV inventory allocation frequently relies on sequential demand hierarchies due to technical constraints in smart TV operating systems and the complexity of real-time bidding in streaming environments. As the CTV ecosystem matures, server-side header bidding alternatives are emerging, but waterfall structures remain embedded in many broadcaster and AVOD (ad-supported video on demand) stacks.
Optimizing a Waterfall When It Cannot Be Replaced
For publishers who cannot move fully to header bidding, waterfall performance depends on accurate tier management:
- Update floors regularly. Static floor prices set months apart allow demand partners to game bids downward. Revisit CPM thresholds weekly against actual clearing data.
- Segment inventory. Run separate waterfall configurations for high-value placements (above-the-fold, video) versus remnant units. Mixing them suppresses revenue for premium positions.
- Monitor pass-through rates. A Tier 1 partner with a consistently high pass-through rate should drop in priority or face renegotiated floor prices. Pass-through above 70 percent signals the partner is not a genuine demand source at that tier.
- Test hybrid setups. Running a header bidding wrapper alongside a waterfall for remaining demand captures real-time competition without a full migration.
Frequently Asked Questions: Waterfall in Advertising
What is the difference between waterfall and header bidding?
The waterfall offers ad impressions to demand sources one at a time in a fixed priority order, while header bidding allows multiple buyers to bid simultaneously in real time. Header bidding consistently produces higher publisher revenue because it finds the actual highest bid rather than the first acceptable one.
Is waterfall advertising still used today?
Yes. Waterfall configurations remain common in mobile in-app advertising and connected TV (CTV), where technical constraints make simultaneous bidding harder to implement. On web display inventory, header bidding has largely replaced the waterfall as the standard approach.
Why did the waterfall method lose popularity?
The waterfall loses publisher revenue because sequential auctions based on historical averages can send impressions to lower-paying buyers when higher-paying buyers sit idle at lower tiers. Header bidding solved this by opening competition to all buyers at once.
What does floor price mean in a waterfall?
A floor price is the minimum CPM a publisher will accept from a demand partner at a given tier. If a buyer’s bid falls below the floor, the impression passes to the next tier. In a waterfall, floor prices must be updated regularly to reflect actual market rates rather than historical averages.
What is daisy chaining in advertising?
Daisy chaining is another name for the waterfall method, describing how ad impressions pass through a chain of demand sources in sequence. Each link in the chain gets one opportunity to fill the impression before it moves to the next.
Related Terms
Understanding the waterfall requires familiarity with the broader programmatic stack. See programmatic advertising, ad exchange, floor price, and demand-side platform for foundational context.