Advertising Waterfall

What is Advertising Waterfall?

Advertising Waterfall is a sequential ad serving method in which a publisher ranks ad networks by expected yield and queries them one at a time until an ad slot is filled. Each ad request starts at the highest-priority network and cascades down to lower-ranked networks only if the ones above cannot fill the slot. This approach dominated mobile and web ad monetization before the rise of in-app bidding and programmatic real-time auctions.

How it works

The waterfall system operates by assigning a priority rank to each ad network based on historical performance metrics such as eCPM, fill rate, and past revenue contribution. When a user triggers an ad request, the system follows these steps:

Step 1: Priority Ranking

Publishers configure a ranked list of ad networks inside a mediation platform. Networks with historically higher eCPMs are placed at the top. This ranking is set manually and updated periodically based on observed performance.

Step 2: Sequential Querying

The mediation platform sends the ad request to the top-ranked network first. If that network returns an ad, the slot is filled and the process ends. If the network cannot fill the slot, the request passes down to the next network in the hierarchy.

Step 3: Cascading Until Fill

This cascade continues through each network in descending priority until an ad is served or all networks have been exhausted. If no network fills the slot, the inventory goes unfilled.

Step 4: Performance-Based Adjustment

Publishers and their ad ops teams regularly review network performance and manually reorder the waterfall to reflect current eCPM trends. This ongoing adjustment is necessary because static rankings can quickly become outdated as market conditions shift.

Waterfall vs. In-App Bidding

The waterfall model contrasts directly with in-app bidding, also called in-app header bidding. In-app bidding allows all networks to submit real-time bids simultaneously, with the highest bid winning the impression. This parallel auction eliminates the sequential delay inherent to the waterfall and often yields higher revenue for publishers. Because of these advantages, in-app bidding has largely replaced the waterfall as the preferred monetization architecture, though some publishers still use hybrid setups.

Why it matters

The advertising waterfall shaped how publishers monetized digital and mobile inventory for well over a decade and remains relevant in contexts where real-time bidding infrastructure is not yet fully adopted. Understanding the waterfall model matters for several reasons.

First, many mediation platforms still support waterfall configurations alongside in-app bidding, and publishers running hybrid setups need to understand how the two interact. Networks that do not support real-time bidding are still integrated via waterfall logic.

Second, the waterfall's limitations directly explain why in-app bidding was developed. The sequential nature of the waterfall introduces latency, as each network timeout adds delay before the next is queried. This latency degrades user experience and can increase abandon rates on ad-supported content.

Third, the fixed ranking structure creates a structural revenue ceiling. A lower-ranked network might be willing to pay a higher eCPM for a specific impression, but the waterfall never gives it the opportunity to compete in real time. This means publishers routinely leave revenue on the table.

Fourth, managing a waterfall is operationally intensive. Ad ops teams must continuously monitor eCPM trends, fill rates, and network behavior across multiple integrations, then manually update the priority order. As publisher portfolios scale, this complexity grows significantly.

For mobile app publishers using an MMP like Airbridge, understanding how waterfall-derived ad revenue connects to downstream user behavior and LTV helps contextualize the business value of individual ad network relationships.

How to Evaluate and Optimize an Advertising Waterfall

Publishers still operating a waterfall setup can follow these steps to maximize its effectiveness while evaluating whether a transition to in-app bidding is warranted.

Audit Current Network Performance

Pull eCPM, fill rate, and revenue data for each ad network over a rolling 30-day window. Identify networks whose eCPM rankings no longer reflect their actual performance. Networks that consistently underperform relative to their rank should be moved down or replaced.

Reorder Based on Current eCPM

Update the waterfall hierarchy to reflect current eCPM averages rather than historical assumptions. Prioritize networks that deliver the highest effective revenue per thousand impressions, not those that simply have the highest nominal CPM floors.

Set Timeout Thresholds

Configure per-network timeout values to limit how long the system waits for a fill before cascading down. Shorter timeouts reduce latency but may miss slower networks with high fill rates. A timeout range of 3 to 5 seconds per network is a common starting point, adjusted based on observed network response times.

Monitor Fill Rate at Each Tier

A network with a high eCPM but a low fill rate creates a bottleneck. If the top-ranked network fills only a small portion of requests, the majority of traffic cascades further down the waterfall and incurs compounding latency. Consider splitting traffic or lowering that network's position if its fill rate is consistently poor.

Evaluate a Hybrid or Full In-App Bidding Transition

Compare revenue outcomes from your waterfall with those from networks that support real-time bidding. Most major mediation platforms allow parallel operation of waterfall and bidding demand sources. Running a controlled comparison across equivalent traffic segments provides a clear revenue differential.

Use a Mediation Platform for Centralized Management

Managing multiple network integrations directly is impractical at scale. A mediation platform centralizes reporting, priority management, and network SDK integrations, reducing the manual burden on ad ops teams and enabling faster optimization cycles.

Related concepts