The bottom line
Mature apps raise eCPM by making more ad networks compete for each user, using mediation tuned by format, device and geography. At this scale, small gains in each segment add up to large revenue jumps.
Key takeaways
- Past performance split by format, device and geography is the base for every decision. 2:30
- Most networks sell only where their sales teams are. AppLovin does very well in the U.S. but may not give the best eCPM in India.
- Rewarded video is a format where Unity is reported to do very well.
- Running bidding and waterfall together fills the gaps that either one leaves alone. 5:36
- Not every partner is available on bidding, so bidding alone leaves demand out.
- Group networks with similar eCPMs, then group them by geography, format and device.
- Small tweaks in each segment bring big revenue gains at scale. 6:37
- Seasonality in network CPMs cannot be predicted, so test small partners on a control group of the audience to spot upticks.
- Developers have seen massive eCPM spikes just from getting this strategy in place.
- The ad experience should match each country's market. 8:17
- In the U.S., where rewarded demand is high, cut interstitials and focus on rewarded.
- In high-volume, low-eCPM markets like India, Brazil and Vietnam, a heavier ad load fits what users already see.
- Three device buckets of high, mid and low end are granular enough. 10:08
- Lists of devices for each tier are easy to find online, and AdMob provides them.
Increasing your competition means that for every ad request, every user that you have, you're making multiple people compete for it.
Sid, Aproda
Chapter summaries
0:00 Where a mature app stands
A mature app has already solved launch and growth, so the open question is how to earn more from users it already has.
- A mature app has scale, a steady product cycle and a mediation setup already running. 0:45
- It has over a million downloads and organic users.
- Updates are planned and scheduled.
- Part one covered launch, where the goal is product market fit and there is no mediation.
- Part two covered growth and the choice between waterfall and bidding.
2:10 Build the baseline, then talk to partners
Higher revenue per user starts with studying past data and learning what each partner can offer.
- A baseline needs historical data and a team that reads it regularly. 2:54
- Earlier mediation experiments supply the data.
- Analytics work grows as the data grows.
- A dedicated team has to watch it on a regular basis.
- Networks behave like sales teams, so eCPM is highest where their salespeople sell.
- Next, shortlist relevant partners and ask each what it can offer. 5:00
- Describe your audience, formats, devices and geographies to each partner.
- Ask what eCPMs each can work with.
- Each partner answers with what works for it and which audiences it has.
- Partners also describe the campaigns they run, which builds your own knowledge.
5:56 Setting up segments and testing seasonality
Segmenting takes work up front, and a control group is the only way to catch network seasonality.
- The setup work pays off because every later optimization builds on it. 6:21
- Once the groups exist, each small segment can be checked for what works and what doesn't.
- The speaker admits it looks like a lot of work, but says everything else then starts falling into place.
- Mediation and bidding are set up so each network is read by geo, format and device.
- Network seasonality has to be measured because it cannot be forecast. 6:51
- Each network has months of very high CPM and months of very low CPM.
- A sales team behind each network drives this, so "March beats April" can't be assumed.
- When a small partner on the control group spikes, move it up in the waterfall.
- Bid optimization and direct deals through Google Ad Manager or AdX are a separate topic.
8:17 Ad experience by country and device
Ad load should follow each market's habits, and device tuning needs only a few broad buckets.
- Country-level ad experience matters most when two or three countries carry the audience. 8:23
- It fits cases where those countries bring about 70 to 80 percent of users.
- Study each market and what its competitors do, then build the experience to match.
- U.S. users face rivals with deeper pockets and clear ad experiences.
- In India, Brazil and Vietnam, volumes are huge and eCPMs low, so heavier ads are tolerated.
- Device tuning works with broad buckets, not individual models. 10:08
- There is no need to separate a Pixel from a Moto.
- The goal is a few larger buckets that are easy to operate.
- AdMob's lists help sort devices into the three tiers.
In the growth library
Sid and Steve P. Young, Sid, ad monetization expert at Aproda (Approach); Steve P. Young, host of App Masters. Mature apps with hundreds of thousands of monthly users can raise AdMob eCPM by segmenting performance by format, device and geography, then running bidding and waterfall together. It also covers testing ad network seasonality with control groups and tuning ad experience per country, which Sid says have produced massive eCPM spikes for developers.
More from
App Masters · The whole library, and ask it anything