3. PERSONAL NEWS FEED
ENABLING A FULLY PERSONALIZED EXPERIENCE
VIA IOS AND ANDROID APP
The ad is shown in a LIVE realistic personalized
context which results in a lifelike reception and valid
insights
IN LIVE
CONTEXT
The new user guidance feature makes sure that
respondents find and watch the ad by scrolling the ad
into view and showing context sensitive instructions
GUARANTEED
EXPOSURE
Support for video and static ads on Facebook,
YouTube, Instagram, Twitter, LinkedIn on
desktop/iOS/Android (more planned)
MULTI-PLATFORM/MULTI-
FORMAT
4. Start in mobile
browser
Link to App
Store
In app start
instructions
Login to live
platform
Test ad is injected
live
ENABLING A FULLY PERSONALIZED AD EXPERIENCE ON IOS AND ANDROID
PERSONAL NEWS FEED
5. NO APPLICATION DOWNLOAD NECESSARY
Respondents do not need to download any app or
extension but use their default device browser. This
maximizes acceptance and reduces panel costs
MAXIMUM PANEL
ACCEPTANCE
The ad is shown in a realistic context with targeted
platform content without the need for the respondents
to remember their login
AUTHENTIC NEWS FEED
CONTENT
Support for video and static ads on Facebook,
YouTube on iOS/Android (more planned)
MULTI-PLATFORM/MULTI-
FORMAT
PUBLIC NEWS FEED
6. Start in mobile
browser
No app
download
No login
necessary
Original look and
feel
Multiformat
IN BROWSER EXPERIENCE - NO APPLICATION
DOWNLOAD NECESSARY
PUBLIC NEWS FEED
7. THE SYNTHESIS OF AUTHENTICITY AND HIGH
PANEL ACCEPTANCE
+
FULL SCREEN EXPERIENCE
Full live platform experience
PERSONAL NEWS FEED
Test ads between personal content
PANEL ACCEPTANCE
Download and login necessary
IN MOBILE BROWSER
Replication of platform UI
PUBLIC NEWS FEED
Generic targeted platform content
PANEL ACCEPTANCE
No download or login
PERSONAL
MOBILE APP
PUBLIC
BROWSER
TECHNOLOGY COMPARISON
8. MOBILE APPS IN-BROWSER REPLICATION
Description
The eye square mobile apps show the real live platform to the respondents
and inject the test ad live in the content. The test ad is surrounded by
original personal platform content.
Respondent visit an authentic replication of an app or mobile platform in
their mobile browser. The test ad is surrounded by public content.
Installation
requirements
Respondents need to download and use the eye square app from the
Appstore/Play Store to see the test ad.
Test is run in the mobile browser and no app download is required from
respondents. This no hurdles solution usually shows high panel
acceptance.
Account
requirements
Respondents need to login to their personal Facebook account in the eye
square app.
No login required so respondents do not have to remember their login
credentials.
Content
authenticity
Respondents see their personal Facebook newsfeed including private
friends posts as they would normally in the platform app.
Respondents see a generic Facebook newsfeed with public content.
The Facebook content is obtained live from the Facebook API and
targeted by language. This approach can not show private friends posts
in the Facebook news feed.
UI authenticity
The platform interface is shown in full screen mode for a most authentic
experience.
Browser address-bar and controls can overlap the newsfeed depending
on the scroll direction (subject to further development)
Ownership
The eye square apps show the live platform and only injects the test ad into
the user interface.
The whole experience is hosted on eye square servers and can be fully
controlled.
TECHNOLOGY COMPARISON
9. The new user guidance feature makes sure that
respondents find and watch the ad by scrolling it into
view and showing context sensitive instructions
FORCED
EXPOSURE
Ads are always shown in a dynamic, authentic context
which results in a most realistic reception and valid
insights
IN LIVE
CONTEXT
FORCED EXPOSURE IN AUTHENTIC CONTEXT
Let respondents browse the platform as they normally
would and get detailed performance metrics for
injected ads
HIDDEN
TRIAL
HIDDEN TRIAL + FORCED EXPOSURE
11. As target audiences consume digital media across all device classes, valid research needs to mirror these device choices. eye square In Context
Research products are fully device agnostic and therefore feasible for all major device classes, formats and browsers.
Test ads across all device
types
DEVICE AGNOSTIC TESTING
13. Product
Automated In Context Pre-tests on YouTube, Facebook and Twitter
Client
Telecommunications corporation
Project period
May 2015 - ongoing, planned to continue throughout 2016
Test volume
350 ads tested as of September 2015
Test in field
United States
Case Study - Telecommunications
SHORT FACTS
14. SHORT FACTS
Client background
The client, a multinational telecommunications corporation is investing heavily
in digital advertising, specifically to target younger age groups on the digital
platforms YouTube, Facebook, Twitter and Instagram. These platforms
feature unique characteristics with respect to viewing context (personalized
content, personal news- and message feeds) and ad format (e.g. skippable
pre-rolls, auto-playing video ads, click to play videos).
Request
The client aims to optimize their creative design to best fit the platforms
viewing context and ad formats.
For this purpose the client seeks to employ an ad pre-testing solution that
allows to quickly test different versions of ads in a standardized way. The test
reports should allow to pick the best performing version of several ads and
provide directions to optimize creative design.
Case Study - Telecommunications
15. Eye square works in cooperation with a Kantar research company to provide the client with a fully
automated pre-tests solution for Facebook, YouTube and Twitter ads.
The solution allows to present the test ads in the authentic viewing context of the personal news feed of
the survey respondent on Facebook, YouTube and Twitter, just like the ad would normally show if the
campaign was already launched.
The solution supports all relevant ad formats (skippable pre-rolls, auto-playing video ads, click to play
videos, right hand side ads) and captures any response behavioral data during the ad exposure phase.
The test ads are being submitted via an easy to use user interface. After submission the test project
gets fielded in a US research panel. An automatically generated report, based on the respondents
behavioral response during the ad exposure and the respondents survey data is available right after
fielding completion.
The project report is typically available in less the 2 hours after the test ads were submitted.
As of September 2015, 350 Facebook, YouTube and Twitter ads have been tested. The tests are
planned to continue throughout 2016.
Solution
16. In-context testing of Facebook ads
Respondents are directed to the Facebook website. They are asked to log into their own Facebook account and
browse through it like they normally do for 60 seconds.
Eye square’s content replacement software is used to include the test ad so that while respondents browse
through their Facebook page they are exposed to the test ad.
After respondents have browsed through their Facebook page they are then directed to look at the ad
specifically and are finally asked the ad diagnostic questions.
In context testing of YouTube pre roll video
Respondents are directed to the live YouTube page and they may pick any video they like to see. Before the
chosen video is shown, the test ad is played as a pre-roll ad and respondents have the opportunity to skip the ad
after 5 seconds just like they have in reality.
After seeing the test ad consumers watch the content video they chose. Finally they are exposed to the ad in full
without being able to skip and asked the ad diagnostic questions.
Test-Setting
17. The eye square – In Context testing software allows ad testing
on live websites.
How it works:
• The participants of an online survey are loading the live
website through the eye square server. The eye square server
modifies the website and then inserts the test ad into the
website. The modified live website is only visible to the survey
participant.
• The software can optionally remove other ads and control the
format and placement of the ad. Generally the technology
allows to control parts of the website, while the original
content remains fully functional.
The In Context testing technology works as a standalone
platform that does not require any involvement or integration with
the media platform being tested. Specifically the approach does
not required to integrate an SDK or code snippets into the tested
media platform.
Simulated Volkswagen ad in the live Facebook newsfeed
In Context Testing Technology
18. Behavioral metrics and analysis
RESULTS
The following slides illustrate the behavioral metrics and
analysis that is performed to allow the selection of best
performing version of test ads and to provide directions to
optimize the creative design.
19. The Viewer Retention represents the maintained interest of the
respondents in different parts of the video ad. It is expressed as
the percentage of viewers who kept the ad playing and kept it in
view at any moment of the play time of the video ad.
Any drop of the percentage of viewers is caused by one of the
following user actions:
• Skipping the ad
• Pausing and not resuming the ad
• Scrolling the ad out of view or navigating to a different page
These user actions generally indicate that the viewer has lost
interest in watching the ad, based on what they have seen
before they performed the action.
The example analysis on the left shows a high viewer drop out
during the early product display phase of the ad (marked in
red), whereas the showing of the female testimonial maintains
the interest of the viewers.
Creative optimization can often be achieved by editing and
repositioning scenes that cause high viewer drop out rates.
5secondsfixed
Maintained
interest
Maintained
interest
Dropoutphase
Numberofviewers[%]
Video time [% of duration]
Viewer Retention
Creative optimization based on Viewer Retention
20. Viewer Retention
Numberofviewers[%]
Percent of video viewed
Percentofvideoviewed[%ofvideotime]
Layering the Viewer Retention graphs of several ads allows to
compare the viewers interest over time across ads on a descriptive
level.
In the example above the “YouTube skippable 2” ad shows an early
drop of the viewers interest at about 20% of the ads play time,
whereas the “YouTube skippable 1” ad maintains the initial viewer
share longer, until about 30% of the video time. The drop in viewer
share of “YT skippable 1” levels out at about 40% viewer share,
compared to just 25% viewer share for “YT skippable 2”
The “Percent of video viewed” metric represents the average
duration the viewers kept the ad playing.
It allows to easily compare an ads ability to maintain the viewers
interest over time, based on a single value.
The diagram above shows that the “YT skippable 1” ad got viewed
on average 60% of it’s total duration. In comparison the “YT
skippable 2” was only viewed up to 47% of it’s duration on average.
Success metrics
21. Viewer Retention
Numberofviewers[%]
Background on classification
While “Percent of video viewed” is an easy to use measure of the viewers
interest in the ad, it does not reflect whether critical elements of the ad (e.g.
the brand or product display) have been in view. It’s also often preferable to
work with a measure of performance that is expressed as a rate of viewers (%
of viewers) vs. using an average time across viewers as it’s the case with
“Percent of video viewed”.
A common approach to develop a related metric is to define threshold view
times for what defines a successful ad view vs. an unsuccessful ad view,
based on the ad objective and type or structure of the ad. As a typical example
viewers who saw less then 25% of the ad could be labeled as “unsuccessful”
views and viewers who saw more then 50% of the ad could be labeled as
“successful” views. Numberofrespondents[%]
Viewer share of Percentage of video viewed
The diagram above shows the percentage of respondents who have
viewed the ad up to a specified range of it’s play time.
Based on the classification described before the “YouTube skippable 1” ad
yields a 17% higher success rate then the “YouTube skippable 2” ad
(49% successful views for ad 1 vs. only 32% for ad 2).
Classifying view times to derive success rates
22. Online ad formats typically provide several features that let
users interact with the ad. Each ad feature (e.g. share, like,
make full-screen, mute/un-mute audio) is tracked
individually and listed in the eye square report.
Like adCursor hover
ad area
Un-mute audio Make full-screen
Facebook autopay ad YouTube skippable pre-roll
Twitter click to play
Analysis of interactive elements
23. Interactivity funnel
The interactivity funnel diagram in the eye square report shows 5
interactivity metrics side by side. The 5 metrics are arranged loosely by
their presumed engagement depth.
The diagram above shows that 77% of the respondents scrolled the
“Facebook autoplay” ad into view – which consequently started the video
replay of the ad. 50% of the viewers hovered their mouse cursor over the
ad and 19% consequently clicked the ad (e.g. made it full screen or
enabled audio). Only 11% of the respondents enabled the audio of the
ad.
Numberofrespondents[%]
It’s important to note that interactivity metrics rely strongly
on the specific ad format and should only be compared
within one ad format.
Interactivity funnel metrics