If you are an app publisher, how do you manage multiple networks, fill unsold inventory, and make sure every impression goes to the highest-paying buyer without spending your entire week on ad ops?
Ad mediation is the answer to that. This guide explains what it is, how it actually works, and what it takes to get a mediation setup running well.
What is ad mediation?
Ad mediation is a layer of technology that sits between your app and the multiple ad networks you work with. Instead of calling one network per ad request, your app sends the request to the mediation platform, which then decides which network should serve the ad.
That decision can be based on pre-set priority rules, historical eCPM data, or a real-time auction, depending on the type of mediation you’re using. The goal in all cases is the same: fill as many impressions as possible at the highest rate available.
Without mediation, you’d have to integrate each ad network’s SDK separately, manage fill rates manually, and guess at which network to call first. With it, that logic is handled automatically.
Why app publishers need it
No single ad network has 100% fill at all times. Demand varies by geography, time of day, ad format, and how competitive a given auction is. If you rely on one network and it can’t fill an impression, that impression earns nothing.
Mediation solves this in two ways. First, it gives you fallback options. If your primary network passes on an impression, the mediation platform passes it down the chain to the next network in line. Second, when you use in-app bidding, it creates real competition, which pushes prices up.
The more demand sources competing for your inventory, the less you leave on the table.
For app publishers specifically, mediation also reduces SDK fragmentation. Instead of maintaining separate integrations for every network, you integrate the mediation platform’s SDK once and connect additional networks through adapters. That’s less engineering overhead and fewer compatibility issues to manage.
How ad mediation works
When a user triggers an ad placement in your app such as opening a new level, finishing a session, or reaching a natural break in the content; your app sends an ad request.
Here’s what happens next:
- The request hits the mediation platform. Your app calls the mediation SDK, which takes over the process of finding an ad to serve.
- The platform runs its decision logic. Depending on the setup, it either works through a priority list of networks (waterfall) or sends the request to multiple networks simultaneously for a real-time auction (in-app bidding).
- A winning network is selected. The network that wins either because it’s first in the waterfall with available inventory, or because it submitted the highest bid.
- The ad is served to the user. The creative renders in your app. An impression is counted. Revenue is attributed to the winning network.
- Data flows back. The mediation platform records what happened: which network won, at what eCPM, for which placement and geography. This data feeds back into optimization over time.
The whole sequence happens in under a second from the user’s perspective.
Waterfall vs in-app bidding
These are the two ways mediation platforms decide which network gets an impression.
Waterfall is the older model. You rank your demand partners in a priority list, usually based on historical eCPM performance. When an ad request comes in, the platform calls Network 1 first. If it passes, it moves to Network 2, then Network 3, and so on. The first network that returns a fill wins, regardless of whether a lower-priority network might have paid more in that moment.
The problem with waterfalls is latency and missed revenue. By the time you’ve worked through several levels, you’ve burned time. And because networks bid based on static historical rates rather than real-time demand, you often leave money on the table.
In-app bidding (also called header bidding for mobile) fixes this. All demand partners submit real-time bids simultaneously. The highest bid wins, every time. There’s no guessing, no stale historical data, and no sequential waiting.
Most publishers running a serious monetization setup have moved toward in-app bidding, or use a hybrid approach where direct-sold and guaranteed deals are handled separately from open auction inventory.
What a mediation platform does
Beyond routing ad requests, a mediation platform typically handles:
SDK management: You integrate once. The platform manages connections to each demand partner through lightweight adapters rather than full SDK integrations for each network.
Reporting and analytics: Unified dashboards that show eCPM, fill rate, revenue, and impressions across all networks in one place instead of logging into five separate dashboards.
A/B testing: The ability to test different configurations, floor prices, or demand partner mixes against each other with real traffic splits.
Floor price management: Setting minimum acceptable bid prices by network, geography, or ad format. Getting floors right is one of the highest-leverage optimizations available to a publisher.
Mediation groups: Organizing demand partners into segments by country tier, ad format, or user segment so each group gets the most relevant set of buyers competing for it.
Automated optimization: More advanced platforms use historical performance data to automatically reorder waterfall levels or adjust floors without manual intervention.
Key things to look for in a mediation setup
Not all mediation platforms are equal, and the choices you make here have a direct impact on yield.
Autonomous ad operations: You need a partner that has a system that’s running even during out of business hours. UndrAds has that, read more here.
In-app bidding support: If the platform only supports waterfall, you’re already behind. Real-time bidding across all demand partners is now the baseline expectation for competitive setups.
Granular reporting: You need to be able to slice performance by placement, format, country, OS, and network. Without that level of visibility, you can’t identify what’s underperforming or where to apply pressure.
Floor price control: Static floors that you set once and forget are not enough. Look for dynamic floor tools that adjust based on demand signals, or at minimum give you easy controls to test different floor levels.
SDK stability and update cadence: A mediation SDK that’s buggy or slow to update when networks push changes creates engineering overhead and missed revenue during downtime windows.
Common mediation mistakes
Running too few demand partners. One or two networks is not mediation, it’s just fallback. The whole point is competition. If you have fewer than five to six active demand sources, you’re not getting the benefit.
Setting floors too high and forgetting about them. Aggressive floors reduce fill rate. If unsold impressions outnumber the revenue gain from higher-paying fills, the floor is hurting you. Test floors against enough data before treating them as fixed.
Ignoring waterfall order decay. In a waterfall setup, if you set your priority order based on last quarter’s eCPMs and never update it, you’ll be routing traffic to networks that no longer win at those rates. Waterfall positions need regular review.
Treating all geographies the same. Tier 1 traffic (US, UK, AU, CA) deserves a different mediation configuration than Tier 3 traffic. Demand partner availability and bid density vary significantly by region. Grouping everything together means your Tier 1 inventory isn’t being competed for properly, and your Tier 3 inventory may have unreachable floor prices.
Not testing ad format mix. Mediation isn’t just about which network wins. The format you’re offering matters too. If you’re only mediating banners, you’re running your auction on the format with the lowest ceiling. Adding rewarded video or interstitials to the mix and mediating across those separately tends to move overall yield more than any network-level optimization.
Assuming the platform optimizes itself. Even platforms with automated optimization need attention. Check your dashboards weekly. Demand partner health changes. New networks launch. Floors need recalibration. A set-and-forget approach slowly erodes performance.
UndrAds works with app publishers to help them increase their ad revenue by automating ad operations. Talk to us to learn more.
