Digital Video Advertising Research
Insights from a Universe of Branding Opportunities
Q4, 2011 Update
Includes data from
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Research
Table of Contents
01
Company Background2
Methodology2
Paid Media: Video Advertising
Why Video Advertising?
Best Medium for Driving Branding Goals3
Going “Viral” Rare; Paid Media Increasingly Common 4
Why Real-Time Media Buying?
Pre-Roll is Catching Up with Display in Terms of
Real-Time Advertising Inventory Available in the U.S.
5
Inventory Snapshot 5
Same Sites; Very Different Prices6
Data & Trends
Creative: Custom or Repurposed TV? 8
What’s Available?10
In-Stream: Where It Works, Where it Falls Short12
Comparisons to Social Video12
Tradeoffs by Ad Format13
A Day(Part) in the Life of Digital Video Ads
15
Maximizing Clicks, Purchases Much Different18
U.S. Video Ad CPMs Steady, Rising Slightly19
Owned Media: Video on Brand Sites
How are Videos Discovered?20
Should I Allow Embeds?20
Rebuffering Nightmares21
Earned Media: YouTube and Beyond
Videos Live Fast, Die Young22
Select Media Company Trends
Peak Viewing Times23
Broadcaster Completion Rates: Full-Length Episodes23
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Overview
02
Company Background
TubeMogul is the only video marketing company built for branding. By integrating real-time media buying, ad serving,
targeting, optimization and brand measurement into its PlayTime platform, TubeMogul simplifies the delivery of video
ads and maximizes the impact of every dollar spend by brand marketers. Over half of the Fortune 500 use TubeMogul to
simplify the delivery of video ads in any format, optimize the impact of their brand message and maximize the effect of
every dollar they spend.
Founded in 2006, TubeMogul is based in Emeryville, CA with offices in New York, London, Chicago, Los Angeles, Toronto
and Sydney.
Methodology
As a real-time media buying platform for video advertising, TubeMogul processes billions of data points on ad spots
available for bidding on a given day across pre-roll, display and other video ad formats. In all, these marketplaces are
bigger than the New York Stock Exchange in terms of daily transactions. Data for this report leverages that data, as well
as TubeMogul’s unique history in video measurement, spanning paid, earned and owned media. Additional video data is
tracked using APIs or crawlers.
More specific methodology is outlined within the individual sections of the report. Additional data is provided by
Brightcove, DynamicLogic and Knowledge Networks.
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
03
Best Medium for Driving Branding Goals
To brand advertisers, video presents an advertising format that is superior in effectiveness to other mediums, even
stacking up well against traditional television buys. Leveraging brand survey benchmarks provided by Dynamic Logic
and TubeMogul’s own numbers for pre-roll video ads from Knowledge Networks, a clear picture emerges: while
television remains the best medium for message association, TubeMogul outperforms TV on favorability and purchase
intent. Intuitively, display advertising does not perform as strongly, supporting suggestions it is mainly a direct-response
medium.
Average Brand Lift
12%
Television
Display
10%
Online Video: General
TubeMogul Platform
8%
6%
11.6%
9.3%
4%
2%
4.3%
3.6%
3.2%
3.6%
4.2%
3.6%
2.2% 2.2%
1.0%
s
es
en
ar
Aw
ge
sa
es tion
M ca
so
As
2.5%
0.7%
0.7%
0.7% 1.1%
d
an
Br ility
b
ra
vo
Fa
n/
tio t
ra nten
e
d I
si
n
Co
se
a
rch
Pu
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
04
Going “Viral” Rare; Paid Media Increasingly Necessary
Fewer than 1% of videos ever exceed one million views, according to our research, which took a random sample of
YouTube videos to compare all-time views.
YouTube Videos by Number of Views
33.3
30%
% of All YouTube Videos
25%
22.8
20%
15%
10%
8.6
10.4
10.2
6.6
5%
5.6
2.3
0%
B
0
w
elo
10
0
50
0-
10
1K
1-
50
.5K
-2
1K
K
5
K-
2.5
0K
-1
5K
K
00
-1
0K
1
0K
10
-<
0.4
on
illi
1m
on
illi
>
1m
Even getting featured on YouTube’s homepage yields fewer views than it used to, despite the site’s meteoric growth
(or perhaps because of it). Videos in the “Featured” or “Spotlight” spots on YouTube’s homepage average 86,100 views
per day that they are featured. That represents a 28.2% reduction compared to the same measurement in 2008, when
videos featured averaged 119,864 views per day.
The result is that paid media is now necessary to get videos watched, even in viral video campaigns.
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
05
Why Real-Time Media Buying?
Conventional wisdom assumes that video ad exchanges are comprised primarily of low-end inventory. That assumption,
often taken for granted, is changing.
Pre-Roll Is Catching Up with Display in Terms of Real-Time Advertising Inventory
Available in the U.S.
While it is widely known that display volume is large due to a glut in inventory, a less known fact is that pre-roll video
advertising is steadily catching up. Aggregating across the major exchanges and sell-side platforms, from Doubleclick
(which includes YouTube pre-roll) to Liverail and beyond and controlling for inventory pushed live due to new company
partnerships, TubeMogul analyzed average growth in daily volume of ad spots available in the U.S.
Pre-roll volume grew by an average of 34.9% per month for the past three months, far outpacing display advertising’s
7.8% monthly growth rate. In November, TubeMogul saw an average of over 213 million auctions per day.
Inventory Snapshot
Running an internal analysis of inventory available for real-time media buying across a five-day sample of over 200
million streams and cross-checking against comScore’s top 100 publishers in Video Metrix, we discovered a few
interesting facts.
• 54 out of 100 companies listed in the comScore 100 have inventory available for real-time bidding on the top exchanges and supply-side platforms.
• 29.1% of pre-roll impressions available for real-time buying are from comScore 100 publishers.
• YouTube, despite its massive reach, only accounts for slightly over one-fifth (20.0%) of comScore
100 volume. Other sources, including top media companies, constitute a majority of sites.
Real-Time Biddable Pre-Roll Inventory
29%
comScore 100 (Video Metrix)
71%
Other Sites
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
Why might this be happening? Publisher-direct sales, often involving faxes and IOs, are not always perfect. Unpredicted
viewership spikes can mean unsold inventory. Also, top exchanges and supply-side platforms are increasingly giving
publishers more control over how they monetize, letting them set minimum CPMs, parameters on who gets access, line-item
controls and much more, freeing up available inventory.
This is not meant to imply that premium video inventory is not scarce, or that video is in any way analogous to display,
where a glut in inventory led to more rapid adoption of exchanges. Recent discussions about private exchanges, and the
role they could play in the marketplace, indicate that many top-tier publishers want additional tools to maximize revenue
before they will be willing to drastically increase the inventory onto real-time exchanges and platforms.
Same Sites; Very Different Prices.
In display advertising, it is commonplace for publishers to plug into more than one automated monetization channel to
liquidate unsold inventory. How prevalent is this in video, and what does it mean for brand advertisers?
TubeMogul examined pre-roll inventory across its breadth of supply sources, including the Web’s top private networks,
platforms and exchanges. The sample spans 266.7 million auctions for ad space over a five day period. From there,
overlapping sites were reviewed and prices of winning auctions compared by both sites and volume.
To control for differences in price by audience demographic, only campaigns without third-party data targeting were
used. Not included in the scope of this study is to what extent publishers distribute different inventory slices to different
liquidation sources, and how natural marketplace fluctuations contribute to pricing disparities over time, making this
more of a market barometer than “buying guide.”
The bulk of pre-roll volume available for real-time media buying comes from the small percentage of publishers (13.7%)
that monetize on multiple platforms.
Sites Monetizing on >1 Platform
80
70
60
All Sites
50
comScore 100
77.6%
40
30
20
10
28.7%
9.3%
0
Sh
all
er s
Ov List
f
e o Site
ar
2.4%
all
er e
Ov lum
f
eo o
ar n V
Sh ctio
u
A
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
06
Digital Video Advertising Best Practices
Paid Media: Video Advertising
By volume, then, most pre-roll is on sites that monetize on multiple platforms. But why would it matter where duplicated
sites were purchased, so long as each ad had a measurable impact and ran in a safe environment?
Where the exact same sites are purchased has a sizable impact on price. CPMs differ by $6.10, on average. This number
is lower for comScore 100 publishers, likely due to the fact that many of these sites have higher average CPMs and more
competitive bidding.
Partnership limitations restrict what can be revealed by inventory source, but it is worth noting that no single platform
stands out as more or less expensive than another -- price varies both within and between platforms. The broader point
is simply that price varies.
Absolute Average Price (CPM) Difference for Identical Sites
$8
$7
$6
$5
$4
$3
$6.10
$5.01
$2
$1
s
ite
lS
Al
00
e 1 x)
or etri
c
S M
m
co ideo
V
(
Here are several conclusions from an advertiser’s point of view, but TubeMogul obviously brings a perspective. We
welcome your feedback.
• Publishers are plugging into multiple supply-side partners to meet their needs and maximize profits – even in video.
• The market is still highly variable and not yet equilibrated – even for identical sites. Demand side platforms (DSPs) plugged into the most video advertising partners are well-positioned to gain visibility into true supply/demand conditions and purchase accordingly.
• A demand side platform might pay for itself purely on the basis of price, given its position to optimize across duplicated sites. Since DSPs charge negotiated, transparent overhead (TubeMogul does, for instance), the calculations are straightforward.
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
07
Digital Video Advertising Best Practices
Paid Media: Video Advertising
08
Creative: Custom or Repurposed TV?
A longstanding question TubeMogul is asked by agencies is: “Should I repurpose my 15 and 30-second television
ads or make custom creative for the Web?” TubeMogul partnered with Dynamic Logic to find out, taking a sample of
hundreds of video ad campaigns, and now have a clearer picture of the tradeoffs involved.
It turns out that there is no one superior production format. Repurposed TV spots typically result in higher impact on
awareness metrics, while made-for-web video content more ably persuades its viewers.
Production Format:
Repurposed TV Ads vs. Made-for-Web Content
5%
4.7
Percent Impacted
Repurposed TV
4.3
4
%
Made-For-Web
3%
2%
2.0
2.2
2.1
1.9
1.6
1.4
1.2
1%
0.8
n
y
er
at
io
ilit
Co
n
sid
Fa
vo
ra
b
In
te
nt
/
ge
sa
Pu
rc
h
as
e
M
es
Br
an
d
As
so
cia
tio
n
ne
ss
Aw
ar
e
Ad
e
lin
On
Ai
d
ed
Br
an
d
Aw
ar
e
ne
ss
0%
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
09
In particular, viewers ages 18-34 exhibit particularly positively impacted purchase intent when exposed to custom video
content.
Age
18-34
6%
5
5.9
35-49
%
50+
4.7
TV
4%
3.7
3%
Web
3.4
3.3
3.0
2.8
2.5
2%
1.7
1.3
1.2
1%
Within each age segment, Made-for-Web
content performs at, or above par for
persuasion metrics
2.0
2.0
1.5
1.4
1.1
1.0
0.7 0.6 0.7
0.5
On
Aw line
ar Ad
en
es
s
Ai
de
d
Aw Br
ar and
en
es
s
0%
Fa Br
vo an
ra d
bi
lit
y
0.4
Repurposed TV out performs
Made-for-Web content in
awareness across age groups
Pu
rc
ha
Co se
ns Int
id en
er t/
at
io
n
Percent Impacted
4.6 4.6
The ability of made-for-web content to impact persuasion metrics is particularly evident among women.
Percent Impacted by Gender
Male
Female
Repurposed TV
Made-for-Web
0.4
0.7
Brand
Favorability
0.5
1.9
0.2
0,4
Purchase Intent /
Consideration
0.2
1.6
0.0
1%
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 2%
©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
10
What’s Available?
Video advertising spans a variety of ad formats, most available for real-time media buying. While this is by no means an
exhaustive list, here is an overview of several available units, to be used for comparisons later.
In-Stream / Pre-Roll: 15, 30 Second Ads (Standard, Ad Selector, Interactive)
In-Display: Any Length (Click to Play, Rollover to Play, Autoplay with “Unmute” Button, Full-Screen Click to Play)
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
Social Media: Any Length (In-Game and In-Application)
11
Mobile: Any Length (In-Stream, In-Display Click to Play)
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
12
In-Stream: Where It Works, Where It Falls Short
In-stream advertising is in high demand, and cost-efficiency is one reason why. According to our research, which took a
sample of 119 million streams from video ad campaigns run through TubeMogul’s media buying platform for video ads,
pre-roll ads deliver a lower effective cost per minute viewed for 15 and 30 second ads than most other video ad types,
including mobile, social media and user-initiated in-display video ads.
30 Second Ads: Spending Necessary to Achieve Same Minutes Viewed
35
30
25
20
30x
15
10
9x
5
0
ll
Ro
ePr
S
ia:
ed
M -App
l
ia In
oc
er
nn y
Ba Pla
In to
r
ve
llo
Ro
In-stream and pre-roll advertising are not without limits, however. While in-stream ads are superior at lifting favorability
and intent metrics, videos run in social media lift awareness and message assocation metrics at higher rates. From
campaigns run in 2011:
Average Brand Lift by Ad Format: In-Stream vs. Social Media
12%
In-Stream
Social Media
10%
8%
6%
4%
2%
0
s
es
en
ar
Aw
ge
sa
es tion
M ca
so
As
d
an
Br ility
b
ra
vo
Fa
e
as t
rch ten
u
P In
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
Using pre-roll exclusively can also mean missing certain viewers. For instance, a marketer buying pre-roll on a newspaper’s site
might miss a broad swath of that site’s audience. According to TubeMogul research in 2010, only 13.1% of unique visitors to a
media company’s homepage end up watching a pre-roll ad on average. By category:
25
Percent of Unique Homepage Visits Resulting in a Completed Pre-Roll
(2010)
20
15
20.4%
10
14.9%
10.5%
5
0
s
r
te
as
c
ad
M
o
Br
s
ine
az
ag
s
er
ap
sp
ew
N
Also, limiting a campaign to 15 and 30-second ads ignores longer-form ads (i.e. branded content). Given the opportunity,
viewers will actually watch longer ads. Again, from campaigns run through TubeMogul’s real-time media buying
platform:
Average Seconds Watched per View by Ad Type: Ads >30 Seconds
100
80
60
85.2
40
45.2
38.6
20
0
32.2
y
Pla
de
out
dA
an
Br
Cl
-to
ick
lay
-P
So
l
cia
M
y
Pla
to
r
ve
llo
ia
ed
Ro
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
13
Digital Video Advertising Best Practices
Paid Media: Video Advertising
14
Other video ad units have unique benefits. For post-view interactions like clicks to brand microsites and social network
sharing, in-banner units outperform pre-roll.
Clicks, Social Network Sharing Rates by Ad Unit (per View)
8%
Brand Site: CTR
7%
Facebook Fan Page: CTR
6%
Facebook Like Rate
5%
4%
3%
2%
1%
0
ic
lay
oP
t
k
Cl
M
ial
ia
ed
c
So
ll
y
Pla
to
r
ve
llo
Ro
ePr
Ro
45.2
38.6 from these findings? The data suggests that a campaign hoping to lift brand metrics across
What should we conclude
the purchase funnel should leverage multiple ad formats to meet its goals.
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
15
A Day(Part) in the Life of Digital Video Ads
While most video ad campaigns are designed to raise awareness and persuade customers, clicks remain a key metric
for many advertisers. Whether the goal is getting people to share a link on Twitter, driving people to have a deeper
conversation on a brand’s Facebook fan page or getting viewers to take an action like filling out a product review or
actually buying a product online, a large contingent of advertisers use clicks as a barometer of specific campaign goals,
especially where survey and other metrics can fall short (i.e. measuring word-of-mouth sharing among friends).
The question then arises for these marketers: you might have reached the right viewer, but did you reach them at the
best time to drive these clicks?
For broader marketers who might not care about click-rates, another question arises: is there an “optimal time to buy” to
minimize cost per minute viewed? Is television primetime more expensive, for example?
Analyzing over 23 million pre-roll ad views in the U.S., TubeMogul utilized its real-time-buying data and campaign
statistics to take a closer look at pricing and click-through-rates by time of day. Campaigns spanned multiple industry
categories, including top brands in the consumer packaged goods, restaurant and retail industries. Both 0:15 and 0:30
second pre-roll ads were included. Every timestamp in the data was localized in order to account for the differing time
zones within the United States.
Overall, completion rates do not differ significantly throughout the day. 0:30 second ads, while mostly stable, dip in the
early morning hours and late afternoon, falling 10% by 3PM from their peak close at midnight.
Completion Rates by Hour of the Day
Completed
Video
Watched
Half
15 Second Ads
30 Second Ads
M
4A
on
No
M
8P
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
16
CPMs for 0:15 and 0:30 second pre-roll ads decline slightly late at night and rebound gradually during the day. However,
the variance over the course of the day is only approximately 10% for both video lengths.
Cost per Thousand Viewers (CPM) by Time of the Day
$12
$6
15 Second Ads
30 Second Ads
M
4A
on
No
M
8P
While CPMs for the two video lengths closely mirror each other over the course of a day, 0:30 second ads consistently
commanded a higher CPM than 0:15 second ads ($9.96 on average, compared to $9.42).
Cost-per-click (CPC) is very similar for both video lengths during the main part of the day (7am-7pm). However, as the
CTR of 0:30 second ads fall in the evening, the CPC becomes more attractive for 0:15 second ads served between 8pm
and midnight.
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
17
Cost-per Click by Time of the Day
$2.50
15 Second Ads
30 Second Ads
$2.00
$1.50
$1.00
$0.50
M
4A
on
No
M
8P
In contrast, click-through-rates follow a distinct pattern during the course of a day, peaking at 5pm and reaching a
trough at 2am. Click-through-rates are slightly higher for 0:30 second ads during working hours, but fall below the rates
for 0:15 second ads during the evening hours.
Click-Rates by Time of Day
1.5%
1.2%
1.0%
0.6%
15 Second Ads
30 Second Ads
M
4A
on
No
M
8P
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
18
Conclusions:
- Click-through-rates on pre-roll video ads follow a distinct pattern according to time of day, peaking at 5pm and reaching a trough at 2am.
- Click-through-rates are slightly higher for 0:30 second ads during working hours, but fall off during the evening hours.
- 30 second ads see 6% higher CPMs than 0:15 second ads and pricing remains relatively steady throughout the day for both video lengths.
Maximizing Clicks, Purchases Much Different
It is important to keep in mind, however, that maximizing clicks or “Likes” and purchase intent are very different goals.
Leveraging its partnership with leading survey vendors, TubeMogul executed brand surveys on five video advertising
campaigns totaling 44.7 million impressions and 4,487 random survey-takers. Four of the five brands are from Fortune
500 companies, and the list reflects the automotive, consumer packaged good (CPG), insurance and restaurant
industries. In terms of format, only 0:15 and 0:30-second pre-roll ads bought via real-time bidding on TubeMogul’s
platform were included. No 3rd-party audience segment data was used; campaigns only leveraged contextual targeting.
Intuitively, the results indicate that purchase consideration and intent metrics tend to increase by the number of times
someone is exposed to an ad, while click-through rates often peak at the first viewing.
15%
Purchase Intent /
Consideration
14.3%
Click-Rates
12%
9%
6%
3.6%
3%
2.0%
1.2%
0
One
1.0%
0.9%
Two
Three
Frequency
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Paid Media: Video Advertising
19
U.S. Video Ad CPMs Steady, Rising Slightly
U.S. Pre-Roll CPMs: comScore 100 Publishers
$12
$10
$8
r
be
S
m
te
ep
er
Oc
b
to
r
be
m
ve
No
As a real-time media buying platform for video advertising that empowers clients to buy across a diverse set
of inventory sources and devices, TubeMogul is in a unique position to look at market trends, including on
price. Think of it as “Yahoo! Finance” for brand advertising online, with this as the sneak preview.
Taking a sample of 29.34 million pre-roll ad streams in the U.S., TubeMogul averaged daily clearing price
per 1000 viewers (CPM). For comScore top 100 Video Metrix publishers, including top media companies and
popular sites like YouTube, CPMs are steady and growing slightly. The average CPM in October was $10.08,
up 5.1% from August’s average of $9.59 but down a few cents from September’s average ($10.12). Expanding
the sample to all sites, CPMs varied little, but were consistently several dollars lower than comScore-ranked
sites.
By category, the most pronounced trend occured in an index of comScore-ranked women’s sites (i.e. Elle,
Glam, Oprah.com), which saw a significant spike, rising to $10.10 in October.
It should be noted that these are the prices from the real-time bidding exchanges, which do not reflect any
overhead charged by TubeMogul or another third party (i.e. a data provider). How is TubeMogul able to
publicly disclose pricing data this way? As noted, the company charges negotiated, transparent overhead to
its platform users.
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Owned Media: Video on Brand Sites
20
How Are Videos Discovered?
Overall, 81.9% of views are discovered via direct traffic or navigation within a brand’s own site. External
sources of traffic are led by Google (58% of all referred streams), followed by Facebook and Yahoo!
Facebook, however, is consistently the fastest-growing external source of views.
Average Monthly Growth in Referral Traffic
20%
15%
18.9%
10%
10.2%
5%
7.1%
0.3%
0
r
tte
i
Tw
!
oo
h
Ya
le
k
oo
og
Go
b
ce
Fa
Should I Allow Embeds?
TubeMogul is often asked by social media teams managing brand microsites whether it is optimal to allow
embeds. Although brands lose some control over where their content is being viewed, the data suggests
the answer is “yes.” For one, embeds boost viewership by 2.6%, on average. Also, viewers discovering brands’
videos offsite tend to watch significantly longer than viewers on the brand site iteself.
Minutes Watched per View: On-site vs. Off-site Embeds
2.0
1.5
1.0
1:42
1:02
0.5
0.0
e
sit
Off
e
sit
On
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Owned Media: Video on Brand Sites
21
Rebuffering Nightmares
While this seems like a minor technical issue, the stakes are high. Why? When viewers encounter a rebuffer
(i.e. the spinning wheel denoting that a video is loading), they click away 81.2% of the time rather than wait
for the content to load.
To calculate this metric, TubeMogul recorded a sample of 1 9 2 , 2 6 8 , 5 6 1 streams from six top video sites and
platforms over a 14-day period, tracking detailed data on video delivery quality. The data comes from
the following top content delivery networks: Akamai, Edgecast and Limelight. The sample is primarily shortform content (i.e. 2 - 1 0 minutes), although long-form TV content is included.
In detail:
• Rebuffers are commonplace, occurring in 6.8% of all streams.
• When encountering a rebuffer, viewers click away 81.2% of the time rather than wait for the video
to re-load.
• Delays in start-time (the amount of time that passes between someone clicking “play” and a stream
starting) were above one second for every network tested.
How Common are Rebuffers?
6.8%
2.5%
>= %
4 St
Re re
bu am
ffe s
rs
>= %
2 St
Re re
bu am
ffe s
rs
of
R
Al ebu
l S ff
tre er
am %
s
1.0%
When you consider that 1 of 15 streams experiences a rebuffer and most people click away when that happens, it’s fair to conclude that far fewer ads are getting clicked (or watched) than advertisers believe.
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Earned Media: YouTube and Beyond
Videos Live Fast, Die Young
22
Way back in June 2008, TubeMogul released research looking at a typical YouTube video’s lifecycle,
breaking out views by number of days passed since a video was uploaded (i.e. one day old, two . . . 90).
While TubeMogul no longer tracks this data, looking back at YouTube’s evolution in the two years it did
remains instructive. Back in 2008, for instance, it took two weeks for a video to get half of its 90-day view
total, whereas in 2010 it only took six days. Similarly, a video in 2008 took 44 days to hit 75% of its total
views, and in 2010 it only 20 days.
As before, hundreds of media companies (i.e. CBS), news outlets (i.e. AP) and YouTube stars (i.e. iJustine)
were included in the sample.
Video Lifecycle: % of Total Views in 90 Days
18%
16%
14%
12%
10%
8%
6%
4%
2%
0%
1
9
18
27
36
45
54
63
72
81
2010
90
2008
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Select Media Company Trends
Peak Viewing Times
23
Overall, a disproportionate share of online video viewing (42.3%) occurs during the eight-hour workday, making online
video one of the best ways to reach people at work. Broken out by category, an interesting trend emerges: broadcasters,
unlike other media companies, have a “primetime” that mirrors broadcast television.
Streams by Hour by Media Category (as % of Daily Views)
7%
Broadcasters
6%
5%
Magazines
Music Videos
Newspapers
4%
3%
2%
1%
M
id
ni
gh
t1A 1A
M M
-2
2A AM
M
3A 3A
M M
-4
4A AM
M
5A 5AM
M
6A 6A
M M
-7
7A AM
M
-8
8A AM
M
9A M 9AM
10 10A
AM
M
11 - 1
AM 1A
M
-N
No oo
on n
1P 1PM
M
2P 2PM
M
3P 3PM
M
4P 4PM
M
5P 5PM
M
-6
6P PM
M
7P 7PM
M
8P 8PM
M
9P 9P
M
M
10 - 10
P
P
M
11 M
PM - 1
- M 1PM
id
ni
gh
t
0%
Broadcaster Completion Rates: Full-Length Episodes
Given their growth and size, broadcasters deserve special attention. While broadcasters often re-post full episodes
online, much of their total title upload count is made up of shorter clips, including episode highlights, previews,
webisodes and other short-form content. However, much of the minutes-watched -- and ad dollars -- are coming from
full-length episodes. So we wondered: do episodes have higher completion rates than clips?
To answer, Brightcove partnered with TubeMogul and took a cross-sample of hundreds of full-length episodes streamed
in the first quarter of 2011, with episode lengths ranging from just over 20 minutes (for half-hour shows on broadcast
television) to over 45 minutes (full-hour shows).
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
Digital Video Advertising Best Practices
Select Media Company Trends
24
The results: completion rates of full episodes are higher than clips. The dropoff at the end of episodes is likely due to
people clicking away during credits.
Broadcaster Completion Rates: Clips vs. Full-Length Episodes
Watches 75%
Completes Video
70.7%
58.8%
63.93%
50.63%
TV
s
ip
Cl
th
ng es
-Le sod
l
l
Fu Epi
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul
www.tubemogul.com | sales@tubemogul.com | 510.653.0684 ©2011 TubeMogul