← All articles日本語で読む →

AdMob waterfall optimization for indie game studios in 2026

May 23, 2026 · 8 min read · Mediation One team

A no-nonsense guide for indie game studios still running AdMob waterfalls in 2026. When to migrate to bidding, when to keep waterfalls, and the four floor-tuning rules that actually move revenue.

AdMob bidding is now the default for most setups in 2026, but a non-trivial number of indie studios still run classic waterfalls — either because they started before bidding was solid, or because they have a custom direct-deal layer that doesn't fit neatly into a bidding-only stack. This article is for those studios.

Should you migrate to bidding at all?

Three signals say yes:

Three signals say wait:

The four floor-tuning rules that actually move revenue

Most indie waterfalls leak revenue not because the structure is wrong but because the floors are slightly off in four specific ways:

Rule 1: Floor by country, not just by ad unit

A single floor across all geos is leaving 10–25% on the table. JP, US, KR, DE, AU, UK all clear at different rates. The minimum useful split is three buckets — Tier 1 (US, JP, UK, AU, DE, CA), Tier 2 (most of Western Europe, ROW high-value), Tier 3 (the rest). AdMob lets you set this at the ad unit level.

Rule 2: Floor by format separately

Rewarded video clears at 3–10x banner. The same studio that floors banner at $0.30 and rewarded at $8 globally is probably either over-flooring banner in Tier 3 or under-flooring rewarded in Tier 1. Don't copy floors across formats.

Rule 3: Re-tune every 30 days

Market eCPM drifts. A floor tuned in November will be 5–20% off by February. Put a recurring calendar item for "floor review" at the same cadence as your QA review. If the answer is "no change", you spent 5 minutes — cheap insurance.

Rule 4: Use the 80th percentile, not the average

Floors should be set near the 80th-percentile clearing price for the ad unit, not the average. The point of a floor is to refuse bad fills; setting it at the average causes you to refuse roughly half of all impressions, which crashes fill and blended eCPM. AdMob's "optimized" floor recommendation usually targets the right zone — sanity check it but trust it as a starting point.

Network-order rules of thumb

For a classic waterfall, the order matters less than people think — modern AdMob will re-sort by realized eCPM within a few days anyway. What matters is which networks you actually include. Reasonable defaults:

Don't over-stack. Past 7–8 networks the marginal revenue from network #9 is below the operational cost of monitoring it.

When to actually go bidding

The honest answer is "when you stop spending time managing the waterfall and start wanting to spend time on something else." Bidding shifts the optimization problem from "ordering and pricing" to "monitoring eCPM and intervening when something looks off." That's a better problem to have, and tools like Mediation One are specifically designed for it.

Try the free 60-second audit

Upload a 14-day AdMob, MAX or LevelPlay CSV and we'll email you the diagnosis. No SDK, no signup.

Run my free audit