November 21, 2012
MoPub, a real-time bidding exchange for mobile advertising, recently published a report on their marketplace for the third quarter which shows some strong trends. The effective cost per thousand impressions (eCPM) was steady or rising across platforms through the last quarter. Android eCPM was rising much faster than for iOS. This is in sharp contrast to the rather dramatic decline in eCPM for smartphones between Q2 and Q3 evident in Opera’s reports, with Android falling faster than iOS.
To put the the MoPub data in context, Opera has multiple mobile ad offerings and their data aggregates different sources of mobile advertising. Some sources have high eCPM and lower fill rates (so an ad is not always available when requested) whilst others have low eCPM but close to 100% fill rates. MoPub’s numbers put them in the latter category, for example, their eCPM for iPhone was just under $0.67 in Q3 while Opera’s was $1.48. MoPub also saw slight declines in eCPM between Q2 and Q3 although nothing like Opera’s 48% and 58% declines for iPhone and Android respectively. Much of the quarterly drop would be expected due to seasonality in mobile advertising spend. Advertisers see the highest conversion ratios in the summer and highest click-through rates around the Christmas holiday season; they spend accordingly.
It’s tempting to speculate that for lower value, higher volume advertising, there is less seasonal variation and Android is gaining on iOS in terms of what advertisers are willing to spend; while for the higher end of the market, iOS remains strong favourite with advertisers but their spending is highly seasonal. The seasonal variation would then mirror the difference in very short term fluctuations in eCPM reported on by Opera:
“High paying ad networks’ eCPM rates fluctuate radically (as much as 370%), while ad networks with lower eCPMs (and high fill rates) fluctuate at lower levels (but still as much as 43%).”
However, with only two quarters of incomplete data from these providers it’s too early to draw any firm conclusions.
The other strong inconsistency between the two providers is for iPad eCPMs. For Opera the iPad went against the seasonal trend and showed increased eCPM in Q3, while MoPub saw the opposite with much larger declines in iPad eCPM, particularly for full screen interstitial ads. The only clear conclusion from this is that developers should not rely on a single ad provider if they want to maximize their ad revenues.
It remains to be seen whether the iPad data will be mirrored for Android tablets. Opera don’t currently split Android tablets out and MoPub only shows a very slight improvement in eCPM over Android phones. We’ll have to wait for more data as the Android tablet market evolves. It will also be interesting to watch how the iPad Mini impacts the iPad rates, or whether it gets split out and reported separately.
What we can see from both providers is that there is stronger demand from advertisers (and thus higher payouts) if more demographic data is provided by the app publisher and also higher rates for rich media ads. Going forward there will likely be increasing demand for more data about users to improve targeting. With the mobile ad industry is currently self-regulating, developers will need to balance privacy concerns with improving their ad revenues. Apple’s highly controlled ecosystem is forcing ad networks to at least give users an opt-out, which Android is still very much open to abuse. Similarly, whilst rich media ads may be more lucrative, they are also more intrusive. A higher rate from a single ad impression is not a good deal if the user doesn’t open the app again. Developers should evaluate the impact on user experience carefully before including them.
Recent Posts
December 17, 2024
What’s Cooking in the 29th edition of Developer Nation survey: A Letter to Our Community
See post
December 17, 2024
The Intersection of AI and APIs: How Technology Enhances Business Operations
See post
December 17, 2024
Preventing Human Error in Development: Essential Tools and Strategies for Error-Free Code
See post