Professional Documents
Culture Documents
April 2013
NEXAGE.COM
Executive Summary
The Q4 Nexage Analytics Report focused on the central measure of mobile advertising health and maturity: market liquidity. Building on our Q4 report, we wanted to understand how value drivers were affecting the market. We felt that the combination of more aggressive mobile strategies from premium publishers, the positive effect of richer and more precise targeting data, a broad base of mobile demand, and the early impact of brand spend would not just grow the market, but shift the market. In essence, we believe the 2013 version of mobile advertising would be fundamentally differentand fundamentally more valuablethan the 2011 and 1H 2012 versions. This report indicates that we are making real and measurable progress:
We reduced the bounce-back periodthe period to which we return to growth after the traditional Q1 down-cycleby a full ve months.
We identied and analyzed attributeschannel, location, app and mobile web, platform, and device IDthat were contributing to the shift in market value.
We gained more clarity into consumers mobile habits and how advertisers and buyers are responding.
The net outcome is that value drivers are having a notable and positive effect on the market. We welcome you to read on for a closer analysis.
Page 1
www.nexage.com
The value shift driven by the combination of higher value audience and broad base (and growing) demand.
The traditional advertising business cycle typically shows a signicant uptick in spend during the holidays followed by a down-cycle in Q1.
We compared Nov 2011-Feb 2012 and Nov 2012-Feb 2013 to understand how they differed. The 2012-2013 cycle showed a more substantial and more sustained holiday spend, a reduced down-cycle, and a sharp spike for the Super Bowl.
Spend
Nov
Dec
Jan
Feb
We examined the bounce-back period between October and Q1. The bounce-back period illustrates how fast we return to growth (if we use October as the baselinethe last normal month before the holiday period). The results show that the value shift reduced the bounceback period by a full ve months. The accelerated bounce-back tells us that the combination of higher value audience and broad base demand strengthened the market, and further signals that brands are increasingly treating mobile as a critical channel to reach consumers.
In 2012, we returned back to growth in May (relative to October) and realized 171% growth in 2012. In 2013, we return to growth in January, ve months earlier than in 2012. Q1 2012 to Q1 2013 growth was 169%.
Spend
Oct
Nov
Dec
Jan
Page 2
www.nexage.com
SUPER BOWL
Spend
Nov
Dec
Jan
Feb
There was a 180% lift in spend on Super Bowl Sunday 2013 as compared to the prior week. The 180% lift compares with just an 18% lift for the 2012 Super Bowl.
Average Spend
180% lift
Super Bowl
Page 3
www.nexage.com
leaders in both ll rate and eCPM performance. Specically, we drilled down to understand which attributes are driving value:
eCPM
We identied 112 sites and apps that are part of the high performance quadrant based on combined eCPM and ll rate performance. The next pages drill down to specic attributes to provide insights into what is driving value in the market.
Page 4
www.nexage.com
All channels are included in the high performance quadrant. No channel represented more than 23% of the total number of sites and apps. Fill Rate eCPM
We further examined how channels compared to each other in two key measures: eCPM and variance. This comparison measures the spread between the highest and lowest eCPM within a specic channel. The greater the variance, the less consensus buyers had regarding the channels value. The results show that although eCPM was comparable across channelsalmost showing a neutral comparisoneach showed stark differences in the level of eCPM variance. These results afrm that channel is not a primary targeting value, and that data is likely having a more signicant impact on performance.
EXHIBIT 6: eCPM PERFORMANCE & VARIANCE BY CHANNEL
eCPM
Travel
Navigation
Entertainment
Sports
Social
Lifestyle
Games
Utilities
Page 5
www.nexage.com
66% of sites providing locationenabled impressions had above average eCPM. 55% of the sites showed above average ll rate performance; 45% showed below average ll rate performance.
EXHIBIT 8: LAT/LONG 67% of sites providing lat/longenabled impressions had above average eCPM. 50% of the sites showed above average ll rate performance; 50% showed below average ll rate performance.
Results were mixed, but telling. Location- and lat/long-enabled impressions showed the eCPM premiums noted in our Q4 report as well as a mixed ll rate performance. 66% of location- and lat/long enabled sites produced better than average eCPM, but ll rate performance was mixed. This result reveals current market realities: not all buyers are able to ingest and target on location and lat/long parameters, which limits ll rate performance. When buyers are able to ingest location parameters, they will place higher value on those impressions.
Page 6
www.nexage.com
54% of mobile properties in the high performance quadrant were mobile web sites. 46% of mobile properties in the high performance quadrant were applications.
Page 7
www.nexage.com
iOS delivered a 118% premium over Android. The premium remains even when we strip out ll rate; 74% of apps that delivered above average eCPM were iOS apps; 26% were Android apps.
We wanted to understand if the 118% iOS premium changed when we looked only at a sample of publishers that provide both iOS and Android apps. Essentially, we took away any effect from publishers that only provided iOS apps or only provided Android apps. For publishers that deliver both, the 118% iOS premium declines to 62%. Inevitably, we believe the 62% premium is more accurate and useful.
EXHIBIT 11: iOS & ANDROID PREMIUMS iOS Premium The iOS premium declined from 118% to 62% when we only look at publishers that deliver content on both platforms. 38% of the publishers realized a premium for Android, but those premiums were less pronounced. Android Premium
Page 8
www.nexage.com
33% of apps in the high performance quadrant enabled device ID. 55% of apps enabling device ID showed above average eCPM performance. 46% of the apps providing device ID showed above average ll rate performance. eCPM
We then investigated whether there was any performance distinction between iOS and Android, especially as it related to the iOS premium noted earlier. The iOS premium persisted, which gave us insights into how developers view lifetime value (LTV). In aggregate, developers place a higher LTV on iOS and buyers are therefore willing to pay more to acquire them.
EXHIBIT 13: DEVICE ID BY PLATFORM
Fill Rate
iOS Android
iOS apps (that deliver device ID) represented 76% of the apps in the high performance quadrant. Fill rate results were mixed, where 49% of iOS apps delivering device ID showed above average ll rate.
Fill Rate
eCPM
Page 9
www.nexage.com
consumption of news and information across the day. They caught up on the news starting around 7am ET and stayed tuned in throughout the day.
4am
12pm
social channels throughout the day, usage spiked in the late evening presumably as they settled in and shared the days events across their networks.
consumption of sports content that spiked as games were being played. The analyses shows that consumers are using mobile for third screen viewing to keep track of scores, storylines, and stats.
The double spike travel pattern: Consumption of travel content oc4am 12pm 8pm
curred during two specic times of the day. During the lunch hour, consumers presumably booked business travel or ex plored getting away from it all; at night, they converted those dreams to specic plans.
consumers across audience segments have a seemingly endless fascination with mobile games, from the early morning through, well, the early morning. Our data af rms those analyses. The only real decline occurs in the wee hours of the morning 1am to 5am ET,
Page 10
www.nexage.com
A value-based model: Similar to TV advertising, this model suggests that marketers and buyers will identify high-value hours to advertise when they assume higher consumer engagement and response. If this model was dominant, we would see advertising volume hone in on specic hours independent of consumption.
A pacing-centric model: This model suggests that buyers will pace their spend throughout the day with possible increases in the later evening. If this model was dominant, we would see a steady pace in advertising volume throughout most of the day, with a slight uptick in the evening to catch up to any pacing gaps.
An economic-driven model: This model suggests that ads served will generally track ad requests, as buyers increase buying when there is a large volume of impressions (and price pressure increases) and pull back slightly when ad requests decline. If this model was dominant, we would see advertising volume track ad requests. However, we would not see a steady ll rate as ads served would not directly respond to spikes in ad requests.
For every channel, we found that an economicdriven model was the primary model. This showed that timing decisions more heavily considered priceassuming they were able to specically target against the attributes such as audience segment, device ID, or location parameter. With that said, buyers are willing to pay more for attributes that drive valueas noted in this report. We would expect this to change over time as brand spend has a more material effect on how and when ads are delivered.
5 6
7 8 9 10 11 12 13 14 1516 17 18 19 20 21 22 23
Page 11
www.nexage.com
Page 12
www.nexage.com