How I Built an eCPM Diagnosis Dashboard for Mobile Game Ad Revenue

Every small mobile game team eventually runs into the same question:

Why did ad revenue change yesterday?

Was it because eCPM dropped, impressions dropped, fill rate changed, one country performed worse, or one ad source started underperforming?

I am building a small tool called eCPM Bazaar to make that diagnosis easier for mini-game and mobile app developers.

The first version focuses on a simple workflow:

• collect ad monetization report data
• normalize revenue, impressions, eCPM, fill rate, country, placement, and ad source
• compare the change by dimension
• surface possible causes when revenue drops

The current version supports demo mode and a TopOn reporting API integration. The long-term goal is to support more sources such as AdMob, AppLovin MAX, Unity LevelPlay, and CSV exports.

Why this matters

A lot of ad dashboards show numbers, but small developers still have to manually answer questions like:

• Did revenue drop because eCPM fell or because impressions fell?
• Is the issue global or only in one country?
• Is one placement causing most of the loss?
• Did fill rate drop while eCPM stayed stable?
• Is the problem tied to a specific ad source?

For example, if US rewarded video revenue drops 25%, but eCPM is stable and fill rate falls from 78% to 54%, the first thing to check is probably fill, waterfall, floor settings, or platform status rather than pricing.

What I am validating

I am not claiming this will magically increase revenue. I am trying to validate whether a clean diagnosis workflow is useful for small teams that do not have a dedicated data analyst.

If you monetize a game or app with ads, I would love feedback:

1. When revenue drops, what do you check first?
2. Do you care more about eCPM, fill rate, match rate, country mix, or placement performance?
3. Would a simple daily alert be useful if it explained the most likely driver of the change?

If this resonates, leave a comment. I am looking for a few developers with real ad monetization data to test the workflow and improve the diagnosis logic.