Featured article
When I first started working in ad monetization, I was quite surprised by how many basic features were missing from mediations.
Five years later, not much has changed.
Actually, in some cases, it feels like things are moving in the wrong direction, because it’s pretty common to see features deprecated rather than improved. We lost support for things like cross-promotion and direct deals, and were pushed hard toward bidding-only setups.
Not to say that nothing has improved. There has been some progress, but overall, we’re missing a lot of features that would make day-to-day work both more efficient and more profitable.
Some of the missing features are honestly a bit baffling. It’s 2026, and one major mediation still doesn't have a proper activity log to see what changed, when it changed, and who changed it. That’s a pretty wild thing to be missing when multiple people are managing millions in monthly ad revenue and something suddenly goes wrong.
To fill the gaps left by poor mediation tooling, a whole range of external ad monetization tools appeared on the market. They’ve been steadily gaining ground, and it feels like almost everyone you talk to these days is either already using some or planning to start soon.
Many of those tools are genuinely useful. But it’s hard not to look at the situation and think that if mediation platforms had really kept up with how publishers work and what they need, many of those companies wouldn’t even exist.
So what’s missing?
Beyond A/B Testing in Ad Monetization
The clearest example is testing.
If I have 1 million DAU, why is my only real native option still a basic A/B test? And even when I do run one, why is AdMob the only mediation that tells me whether the result is statistically valid or not?
That should be standard, not some standout feature.
But the bigger issue is that even those A/B tests don’t cover most of the things we would actually want to test. In most cases, testing is limited to an ad unit or mediation group level. Anything beyond that and you’re basically on your own.
In an ideal world, we would be able to set up A/B/C/n tests. And we wouldn’t be limited to a narrow, predefined set of variables that mediations allow us to test.
And the lack of those options actively hurts optimization in practice.
Let’s say you want to test a new, lower bid floor.
You set everything up, and the new group shows a clearly winning result. Maybe ARPDAU is up 20%. Great. So you promote the test and expect to see a nice uplift.
And then your ARPDAU barely moves.
What happened?
Well, all these smart algorithms quickly figure out that they can bid for cheaper. It doesn’t really matter that your users are split between test groups if, underneath that, you’re still running on the same ad unit and the same network IDs. Networks just stop bidding as aggressively on your “expensive” control group because they can access similar users more cheaply somewhere else in the setup.
That kind of contamination is common enough that, combined with the fact that some tests cannot even be run within the mediation itself, means that the majority of serious testing needs to be done with external tools.
The Challenges of Multiple Ad Unit Setups
The difficulties get worse once you get into the more complex setups publishers use today.
The move to an almost entirely bidding-only environment pushed ad monetization teams to claw back control through all sorts of creative approaches. Probably the most successful of them has been multiple ad unit setups with bid floors.
It’s an irony of modern ad tech: to make bidding work, we had to recreate the waterfall manually.
Anyone who has worked on one of these knows how tricky it is to get it right.
How many ad units should I run? What’s the optimal delay between different calls? How often do I change the bid floors? How aggressively should they move? There are a lot of decisions to make, and mediation platforms offer very little help in designing this properly. So you do it yourself: test, iterate, patch things together, and eventually, after a few rounds, it starts working well.
And then what?
Now you have a setup where several ad units are effectively being used as one monetization system, but the mediation platform still treats them as separate pieces. Which means you basically lose the ability to run clean tests on the whole thing.
Yes, technically, you can still run tests on the ad unit level. But you have no real idea how that change impacts the other ad units in the chain, which makes the whole exercise a lot less useful.
Ad Quality: The Variable Mediations Still Can't Control
Another serious issue is ad quality, which has become one of the biggest pain points for mobile developers.
Why does it matter if ARPDAU goes up 5% if churn rises enough that total LTV actually gets worse?
The problem is that the tools to deal with this are still very crude.
Want to test whether blocking a certain competitor helps? Want to compare a more conservative ad template against a more aggressive one? Networks still don’t give you a proper way to test any of it. The only comparison you have is before and after, which makes it very hard to separate the effect of the change from everything else that was happening at the same time.
And this is only scratching the surface. There are so many missing features that it’s not even worth trying to make a complete list here. One simple example: MAX by AppLovin, the biggest mediation in the world, still doesn’t have something as simple as app-level fill rate.
Ad Infrastructure Shouldn't Need to Rely on Workarounds
The sad reality is that mediation platforms have not kept up with the operational complexity of modern ad monetization. Publishers are forced to rebuild missing core functionalities with hacks, external tools, and custom solutions. People have simply gotten used to working around mediations. All this seems untenable long-term.
The fact that these patchwork solutions are starting to feel normal is not how mature infrastructure is supposed to work. And for something as central as mediation, it’s hard not to see that as stagnation.
