Mediation And Waterfalls: How App UA Buying Actually Works

The short version

In-app advertising inventory is sold through publishers who run many demand sources at once. Mediation is the layer that decides which network gets to fill each ad slot.

This matters to advertisers because your access to inventory is decided by a process you do not participate in directly. Your creative can lose volume for reasons that have nothing to do with your bid or your targeting.

Waterfall versus in-app bidding

Waterfall. The publisher orders demand sources by historical average revenue and offers the impression to each in turn until one takes it. Ordering is based on past performance, so a network that performed well historically gets first look regardless of what it would pay for this specific impression.

In-app bidding. All sources bid in real time for each impression, and the highest wins. Closer to how programmatic works elsewhere.

Most real setups are a mix, with some sources bidding and others sitting in a waterfall.

Why it affects you as an advertiser

Your effective reach depends on your network's position. A network low in a waterfall sees only impressions the sources above it declined. Same bid, worse inventory.

Creative performance drives delivery, not just efficiency. On networks that rank creative by IPM, a weak IPM directly reduces the volume you can access. This is different from social, where a weak creative mainly costs efficiency. Here it costs availability.

Price signals are indirect. Raising your bid may not increase volume if the constraint is positional rather than price.

The same impression is worth different amounts to different networks. Which is why the same campaign performs differently across networks in ways that are not about your targeting.

What this means practically

IPM is not just a metric here, it is an access mechanism. Improving creative pull increases how much inventory you can reach. This is the strongest argument for creative volume in app UA specifically.

Network diversity is not optional. Because each network's position varies by publisher, running on several is how you reach inventory any single one cannot.

Test creative per network. A creative that wins on one network can underperform on another, because the audiences and the ranking mechanics differ. Do not assume a winner transfers.

Volume ceilings are real and often not about money. When a network stops scaling, more budget frequently does nothing. The constraint is what that network can access, not what you will pay.

What you cannot control

You do not choose your position in a publisher's waterfall, you do not see the full auction, and you cannot verify most of it. Reporting differences between networks and your MMP are normal here for structural reasons rather than because something broke.

The useful posture is to treat network-reported numbers as one input and your MMP as the arbiter.

Practical rule

In app UA, creative quality buys access as well as efficiency. When a network stops scaling, look at creative and network mix before assuming the answer is a higher bid.