Implementing an audience-based cross-platform video strategy across a large media holding company is a long journey – and one that Omnicom Media Group has recently embarked upon. OMG’s advanced TV platform—a component of Omnicom’s Omni people-based precision marketing and insights platform—ties together audience-based TV planning, buying and measurement capabilities across data-driven linear TV, household-addressable and IP-delivered OTT platforms. Jonathan will share the vision, trajectory and some early lessons learned on the early part of this journey.
All Aboard! A Cross-Agency Mission Toward Cross-Platform Audience-Based Video
1. 1
All Aboard!
A Cross-Agency Mission Toward
Cross-Platform Audience-Based
Video Planning & Measurement
MediaPost TV & Video Insider Summit
February 25, 2019
Jonathan Steuer
Chief Research Officer, Omnicom Media Group
jonathan.steuer@omnicommediagroup.com
2. Here’s where we’re headed…
• TV in transition
• The “Unreachables” today
• Stacking the future
• Picking a platform
• Data at the core
• Platform components
• Early case studies
2
Agenda
3. Technology advancements and device proliferation are
changing TV distribution models
3
New Distribution Fragmentation Tech & Data
4. Content attract humans
TV is slowly moving beyond broad demography, evolving how
we both target and measure video delivery
(Past into Present) (Present into Future)
Humans targeted across content
Audiences attracted by content Audiences aggregated based on data
Ads tied to content Content and ads separable
Measured by Nielsen panel proxy Measured by actual delivery
(challenge: translate delivery to impressions)
Content-First Audience-First
Measurement against age/gender demos Measured against precise targets
4
5. “The Unreachables”
According to OMG primary research,
Millenials & Gen-Xers are consuming 30
hours of video content per week, but only
10 hours of this viewership is captured
in the Nielsen linear TV currency
47% of respondents reported zero
viewership of Nielsen-measured linear
TV–hence are “unreachable” from the
viewpoint of Nielsen-measured TV reach
Source: Hearts & Science and OMG RISE Primary Research
Unreachables Research, 2017
5
TV with
antenna
TV with
cable or
satellite
TV with a
streaming
device
Smart TV
app
Desktop/
Laptop
Smartphone
(browser)
Smartphone
(app)
Tablet
Video
game
console
Live TV
Traditional TV:
10 Hours
(What today’s currency of
TV measurement is now
equipped to capture)
TV content on other devices:
5 hours
DVR
Cable/
Satellite
TV
Streaming
Video
Streaming content across all devices:
15 hoursShort form
videos
Hours of Video Consumption in a Typical Week
6. Isolating the the “Unreachables” shows the real declines in linear TV
Linear TV viewing declines are 2-3x steeper among younger viewers than A35+ — 5% annual declines are only the P18-49 average!
6
31.67
24.46
30.63
35.92
23.20
13.89
20.06
29.90
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
P18-49 P18-24 P25-34 P35-49
2012/13 2013/14 2014/15 2015/16 2016/17 2017/18
-27%
Source: Nielsen National TV Toolbox
Linear TV Viewership Trend 2012/13 – 2017/18
(Average Hours per Week)
-43%
-35%
-17%
7. Check the stack: Live
viewership is migrating to OTT
7
• Total viewership of linear content is across
devices is relatively flat (-2% on NBC, shown)
• “Stacking” measurement by network is not
syndicated by Nielsen–TCR is for networks only
(requests out to other networks for their data)
• “Fluidity” and ADUs taken in digital are on the table
• NBCU’s C-flight represents an attempt to measure
and steps in the right direction
• But we need to change the way we plan, buy and
measure
10. OMG explored the Advanced TV landscape in 2017
10
Network Group Offerings
TV network groups enabling audience-based
buying across their portfolios
Inventory holders optimize inventory themselves
and price on a premium basis
Challenging to tie platforms together for unified reporting and
inventory may be self-serving to fulfil yield
Platforms either facilitate agency solutions or provide
client-direct solutions for planning, optimization, and buying
Quality of inventory, data availability, and business
models vary considerably across platforms
Programmatic TV solutions frequently bundle remnant inventory
Platform-as-a-service model recommended with VideoAmp
Advanced TV Platforms
11. OMG Advanced TV Platform powered by and TV Viewership data from
OMG selected VideoAmp as our development
partner to implement Omni Advanced TV
11
Cross-Channel Measurement (Aggregate Reach and Frequency)
Plan & Optimize
Leverage linear TV and digital data to
plan integrated video investments
Allocate
Optimize integrated video investments with
our data and software solutions
Measure
Understand duplication of audiences
between your linear and digital investment
Client Audience/Targeting Data from DMP (1st Party, 3rd Party)
Data-Driven Linear TV IPTV Household Addressable
Broadcast and Cable Network
(Local & National) Inventory
OTT/FEP/CTV Inventory MVPD Inventory
12. The platform combines, cleans, and matches TV viewership data
across ACR and set top box data from multiple sources to
enhance representativeness and accuracy
12
Technology Provider Offering Data Source Pros Cons Status
ACR Technology
and Data provider
Most Vizio Smart
TVs
Collects data from a
universe of 10MM opt-in
TVs
Some skews towards
higher income,
homeowners, and
married (statistically
correctable)
In platform today,
including weighting
and calibration
STB Data provider
Collects from STBs in
1.1 million
households
concentrated in
Florida, California and
Texas
No limitations on data usage.
Data includes TV viewership,
VOD, and DVR for northeast,
west coast, and central Texas
Smaller sample of
households than the
larger cable companies
Commingling in
progress, available Q3
2019
13. Commingled STB and ACR data is unique to our platform
13
Step 1:
Merge common
fields present in
STB and ACR TV
viewership
Step 2:
Determine overlap
of STB and ACR
households
Step 3:
Inform algorithms
that classify zombie
viewing in the STB
data feed with ACR
data
Step 4:
Use STB data to
understand
programming
viewership on
multiple TVs in the
ACR households
Step 5:
Deduplicate
overlapping TVs in
households to
prevent double
counting in analysis
14. How Omni Advanced TV works for our clients
14
1st & 3rd
Party Data
Leverage website, CRM,
sales or 3rd party data
to create targets
Match IDs to
TV HH
Match targets from
online & offline data
to TV Households
Build Behavioral
Indices
Create indices based
on viewership and
online behaviors
Plan Against
Audiences
Create plans to reach
audiences in highest
concentration most
efficiently
Measure
Performance
Analyze media
performance to help
optimize future
campaigns
15. Linear Planner
• Apply any data asset
to linear planning
scenarios to optimize
for linear upfront
planning and
negotiation
Linear Allocation
• Optimize multi-brand
level linear
allocation
API-based TV
Transactions
• Plan & buy linear
scatter TV inventory
from sources directly
connected to
platform
Omni Advanced TV
will enable more
effective planning of
video investment via
a continuous
optimization loop
15
TV Digital
Segments
• Create digital
audiences based
on linear TV ads to
drive incremental
R/F or conquest in
desired channels
Cross-Screen
Measurement
• Measure de-
duplicated R/F
between linear &
digital ad exposure
16. Enables advanced
audience planning
and leverages
algorithmic
optimization based
on data ingestion
to increase the
efficacy and
efficiency of
upfront planning
LinearPlanning
16
TARGET 2
TARGET 1
TARGET 3
17. Advanced Audience Targeting Data
Manages upfront investments across brand and product portfolios
throughout the year, identifying the most efficient distribution
against advanced audiences
17
LinearAllocation
TV Advertising Investment
Brand X
Product E
Network 1
Network 2
Product F
Network 3
Network 4
Product C
Network 5
Network 6
Product D
Network 1
Network 2
Brand O
Network 1
Network 2
Network 3
Product B
Network 3
Network 4
Product A
Network 1
Network 2
Phase 1
File-based allocation reports
by Q4 2018
Phase 2
File-based API ready at end
of Q1 2019
Phase 3
Fully-functioning self-serve
UI by Q2 2019
Brand Y Brand Z$ $ $ $
18. Surfaces available scatter inventory for discovery, and enables
planning / execution against advanced audiences with visualized
plan performance forecasts across networks and dayparts
API-basedTVTransactions
18
Phase 1 - Proof-of-concept testing
of forecasting methodologies and
normalization of inventory sources
Phase 2 - Initial release with file-
based plans with Tier 1 networks
Phase 3 - Enhanced version
incorporating a self-service UI
connected to additional inventory
sources
19. Creates optimal plans to reach audiences at the proper
frequency levels by analyzing cross-screen delivery
19
Cross-ScreenMeasurement
Duplication
1.0M
Linear Reach (HH)
5.5M
Digital Reach (HH)
5.0M
Increase on-target reach
and control frequency
across linear and digital
households to reduce
waste
Understand delivery at the
network, daypart, and
program level
Identify investment
opportunities across
channels
15% of households who saw
a digital ad also saw a TV ad
13% of households who saw
a TV ad also saw a digital ad
20. Measures de-duplicated reach and frequency
20
Cross-ScreenMeasurement
Implement Cross-
Screen Measurement
Report on linear and digital
video to understand reach and
frequency, where to gain
incremental reach, and
exclusive reach by partner.
Reduce Waste
Combine impression,
viewership and ad exposure
data with household-level
brand exposure across all
channels and screens in a
privacy compliant manner.
Optimize Against Effective
Reach & Frequencies
Analyze unified data set to help
determine which audiences,
from both a TV and digital
perspective, are being either
under or over exposed.
Improve Strategy
& Buying
Maximize reach by providing
actionable insights to apply
against future or immediate
campaign investment
opportunities across all TV
and digital media.
24. API-based inventory connectivity enables planning and execution
against advanced audiences across platforms, networks and dayparts
CaseStudy:API-basedTVTransactionPipes
Future
(in progress)
Linear Network
FEP / OTT
MVPDs/vMVPDs/TV Anywhere
Phase 1
Proof-of-concept testing of
forecasting methodologies and
normalization of inventory sources
Phase 2
Initial release with file-based
plans with Tier 1 networks
Phase 3
Enhanced version incorporating a self-
service UI connected to additional inventory
sources
25. OMG is moving!
• Platform implementation continues
• Investment partnerships include data and inventory access
• Widespread training happening now, in preparation for Upfronts
• Client education and ability to compare methods is critical for adoption
• Many / most OMG clients will touch the platform in next 2 months
(a parallel process for this year)
• The normal way in 2020-21?
25
TheJourneyInProgress
Cord-Cutting Snowballs: U.S. Pay-TV Ops Shed 3.2M Subscribers in 2018 – Variety
https://variety.com/2019/biz/news/cord-cutting-2018-accelerate-us-pay-tv-subscribers-1203138404/
In response to technological changes, TV distribution models have changed massively in recent years. Audiences are now highly fragmented across devices and inventory sources, necessitating new data and technology to plan, target, and measure TV delivery, efficiency, and performance. While these changes have challenged traditional linear TV models, the altered industry dynamics have presented an opportunity to leverage new data and technology to bring an audience-based approach to TV.
As a result of the declines in traditional TV viewership, new approaches are needed. In particular, the TV industry needs to migrate from planning against broad demography to an audience-based approach. This will enable teams to reach target audiences across all the content they view, to aggregate those audiences based on data, and to measure delivery against those audiences rather than to rely on content or a broad demographic as a proxy for an audience.
In order to enable audience-based TV, new datasets with more granularity and identifiers that can be matched to 1st party, 2nd party, and 3rd party datasets are needed. Examples of this type of viewership data include Smart TV data collected via ACR (automatic content recognition) technology and set top box data. Here providers such as Inscape (for Smart TV ACR) and Frontier (for Set Top Box data) can be leveraged to power advanced audience solutions.
According to OMG Rise primary research, adults 18-49 – the core demographic that linear TV is transacted on – are now consuming 30 hours of video content per week, but only 10 of those hours are captured in the Nielsen currency. 5 hours is spent watching TV content on devices not measured by Nielsen, and 15 hours are spent viewing streaming content across various devices.
These changes in viewership behavior will continue to exacerbate the TV measurement problem as consumers migrate to viewing via OTT sources rather than traditional TV. The declines in viewership are more drastic among viewers under the age of 35, with viewership in the 18-24 bracket declining 43% and viewership in the 25-34 bracket declining 35%.
Omni is Omnicom’s people-based precision marketing and insights platform, designed to identify and define personalized consumer experiences at scale, in order to drive superior business outcomes for clients.
Omni transforms the way our teams work, collaborate and deliver value by providing a single-view of the consumer from insights development to audience building, channel planning, content inspiration, and message distribution.
Advanced TV Solutions:
As the TV market has evolved, the marketplace has produced two types of solutions. The first type is network group solutions. These are offering from TV network ownership groups and their consortiums. At this point, most of the major networks have their own advanced TV offerings, but the capabilities vary considerably. Most are still powered by Nielsen and have only panel 3rd party datasets such as MRI. The most prominent of these solutions is OpenAP a consortium of Fox, Viacom, Turner, and NBCU. However, there are major drawbacks to all of these solutions. Primarily, none of them provide a holistic view of the market, they only optimize within their participant networks, which severely limits the value of their optimization. In addition, the methodology of optimization underlying these tools is opaque, with no insight into which inventory is selected. In other words, as an agency, we are unable to tell whether spots are assigned to increase reach against an audience we care about or if it has been assigned to improve the network’s yield.
The other set of solutions comes from Advanced TV platforms. There are numerous providers in the space that broadly fit into two categories. Platforms that essentially operate as ad networks (e.g. Adobe/Tube Mogul and Simulmedia), rolling up remnant inventory and selling it non-transparently as premium inventory and platforms that operate as agency partners under a platform-as-a-service model (VideoAmp and 4C Insights). The former are not recommended (obviously), while we have an internal solution powered by one of the latter.
Omnicom’s Advanced TV Solution:
Omnicom has developed an integrated media, data, and software solution, powered by VideoAmp, to enable holistic planning, buying, and measurement across TV and video distribution channels. The platform is built upon granular household-level TV viewership data matched to strategic targets from either the client’s DMP, the Omni Audience Builder, or other sources via the Agile ID. In day-to-day operation, the tool provides strategy, marketing science, and investment teams with a point-and-click interface to evaluate and optimize the performance of a client’s 1st and 3rd party audiences across channels in the video ecosystem.
Viewership Data:
There are 3 types of TV viewership data available in the market – panels, set top boxes, and ACR providers. Ideally we would have a dataset that merges all three sources of TV data, but today this is not available.
Panels recruit individuals record their TV viewing habits through paper diaries, meters, or cameras in their households. Typically, panels will have a recruitment process to ensure that there are enough individuals in a given area or demographic. This allows the data provider to weight their data in a statistically stable way to be representative of the US. However, many panels are relatively small which leads to the data being heavily weighted making it difficult to understand local and niche program viewership. Most panels limit how much data processing and manipulation can be done by their licensees, which forces users of the data to rely on the vendor to correctly analyze the data on their behalf. Examples of panels include Nielsen and ComScore.
Set top box data is collected by multi-channel video programming distributors (MVPDs – e.g. Comcast, Spectrum, AT&T/DirecTV, Verizon Fios). A set-top box is the device that requests from the video server and displays the TV programming for a specific channel and time. It is possible for data to be collected when a STB is turned on and the TV is off, a phenomenon called zombie viewing, which can cause inaccuracies in measurement if not properly accounted for. Depending on the collection methodology of the specific provider, the STB may be able to track linear, VOD, and/or DVR viewership.
Automatic Content Recognition (ACR) enables the capture of images and/or audio fingerprints being played on Smart TVs which can then be matched to content or advertising feeds. ACR captures live TV, VOD & DVR playback, and OTT consumption across a number of different viewing sources. The information that is captured from the TVs is passed back to the device manufacturer along with an IP address. This results in the manufacturers having large-scale (the size of their opted-in footprints) event-level viewership data. Within the ACR landscape only Vizio/Inscape licenses their viewership data for agency use, while others take walled-garden approaches (Samsung) or managed service models (SambaTV, Alphonso) with limited transparency into methodology or the underlying data.
OMG’s data solution will include Frontier STB data before the end of 2018.
Inform ACR Data
ACR data tends to only capture one TV in a household. Cable data will be able to inform us on what the other TVs in the households are watching.
Flexibility of data
Frontier will provide us with data feeds with no conditions attached. We will be able to manipulate and cut the data for any research and client use cases.
Spread of subscribers
Frontier is not exclusively in urban or rural areas, so we will have a sampling from many geographic areas. Skews in demographics can be corrected for, but niche viewership may be limited depending on request.
Viewership Data:
There are 3 types of TV viewership data available in the market – panels, set top boxes, and ACR providers. Ideally we would have a dataset that merges all three sources of TV data, but today this is not available.
Panels recruit individuals record their TV viewing habits through paper diaries, meters, or cameras in their households. Typically, panels will have a recruitment process to ensure that there are enough individuals in a given area or demographic. This allows the data provider to weight their data in a statistically stable way to be representative of the US. However, many panels are relatively small which leads to the data being heavily weighted making it difficult to understand local and niche program viewership. Most panels limit how much data processing and manipulation can be done by their licensees, which forces users of the data to rely on the vendor to correctly analyze the data on their behalf. Examples of panels include Nielsen and ComScore.
Set top box data is collected by multi-channel video programming distributors (MVPDs – e.g. Comcast, Spectrum, AT&T/DirecTV, Verizon Fios). A set-top box is the device that requests from the video server and displays the TV programming for a specific channel and time. It is possible for data to be collected when a STB is turned on and the TV is off, a phenomenon called zombie viewing, which can cause inaccuracies in measurement if not properly accounted for. Depending on the collection methodology of the specific provider, the STB may be able to track linear, VOD, and/or DVR viewership.
Automatic Content Recognition (ACR) enables the capture of images and/or audio fingerprints being played on Smart TVs which can then be matched to content or advertising feeds. ACR captures live TV, VOD & DVR playback, and OTT consumption across a number of different viewing sources. The information that is captured from the TVs is passed back to the device manufacturer along with an IP address. This results in the manufacturers having large-scale (the size of their opted-in footprints) event-level viewership data. Within the ACR landscape only Vizio/Inscape licenses their viewership data for agency use, while others take walled-garden approaches (Samsung) or managed service models (SambaTV, Alphonso) with limited transparency into methodology or the underlying data.
OMG’s data solution will include Frontier STB data before the end of 2018.
Inform ACR Data
ACR data tends to only capture one TV in a household. Cable data will be able to inform us on what the other TVs in the households are watching.
Flexibility of data
Frontier will provide us with data feeds with no conditions attached. We will be able to manipulate and cut the data for any research and client use cases.
Spread of subscribers
Frontier is not exclusively in urban or rural areas, so we will have a sampling from many geographic areas. Skews in demographics can be corrected for, but niche viewership may be limited depending on request.
Optimization of Linear TV Plans:
Once a client’s desired 1st and 3rd party audiences have been identified, these audiences are matched to the viewership data in the platform via the Agile ID. In the platform, the strategy, marketing science, and investment teams collaborate to create scenario plans against the strategic audiences for each of the brands. Since the majority of linear TV is still transacted offline using Nielsen ratings as the currency, the plan outputs provide a side-by-side comparison of the Nielsen target and the strategic audience enabling investment teams to gain an asymmetrical advantage in negotiation with the network. The platform optimizes delivery concentrations down to the program level based on viewing and other behaviors captured in the audience definitions, maximizing gains in efficiency and effectiveness.
After a client has committed to inventory from the networks (during the upfronts and/or in the scatter market), the platform has direct API connectivity to the client’s inventory in MediaOcean to allocate across the brand portfolio. Inventory is allocated according to concentrations of the strategic audience targets for each of the client’s brands as well as delivery of Nielsen targets. Once units have been allocated, the finalized plans for each brand are loaded back into MediaOcean for record-keeping and billing purposes.
Linear TV Upfront Planner
Apply any data asset to linear planning scenarios to optimize for linear upfront planning and negotiation
Linear Allocation
Optimize multi-brand level linear allocation
Linear Scatter TV Buying
Plan & buy linear scatter TV inventory from sources directly connected to platform
Cross-Screen Measurement
Measure de-duplicated R/F between linear & digital ad exposure
TV Digital Segments
Create digital audiences based on linear TV ads to drive incremental R/F or conquest in desired channels
Identify past success and future opportunity
Through layering advanced audiences on historical buys to establish target baselines
Build better upfront plans and smarter scatter plans
Through greater workflow efficiency and audience insights
Allocate and activate on linear and digital with cross-screen intelligence
Through incremental reach discovery and digital extensions of TV plans
Cross-Platform Video Optimization & Performance Measurement:
The same targets used for linear TV targeting can also be pushed directly to OTT, FEP, addressable VOD and digital platforms that enable such granular targeting. Linear TV delivery information against strategic targets can be used to create digital targets that optimize for incremental reach beyond the audiences delivered via linear TV. Household-level TV delivery data is also combined with device-level digital platform delivery data to create unified reach and frequency reporting across all video sources. Such combined delivery data enables the ability to develop advanced performance metrics.
Custom 1st party DMP segments for planning identified
Map custom MRI segments to TV viewership data at the airings level
Planning Segments Driven Based on Combination of Clusters from Actual Purchase Behavior
Build custom network indexing across 3 segments: McCafe, Breakfast, Classics, Value
Utilize Custom indexing plus other factors for optimizations
Run allocator taking only GRP goals into account. System allocates inventory to brands in flight to tie back to their GRP goals. Take snapshot of the results
Manually tweak allocations based on vendors that over index on estimated segment ratings and then on segment composition. Take snapshot of new results
Inventory allocation is communicated to vendor for air
Brand lift can be identified in strategic demand spaces
First, optimize towards segment rating indexes that are higher than the network average index
Second, optimize towards segment % Comp Indexes that are higher than the network average index
Methodology
Calculated average ratings / indexes for network, dayparts and or programming against each CRM segments
Scored and ranked networks, dayparts and programming against the indexes
Applied McDonald’s actual CPMs for accuracy
Utilized both CRM indexing and actual costs to develop the recommended plans
Proposed scenario offers ability to deliver $6.8MM in cash value while over-indexing GRPs
Opportunity to remove 5% of total budget while still over-delivering A1849 GRPs and optimizing plan on McDonalds segments vs. baseline
OMG Advanced TV Platform
Get ready for Upfront planning
Prepare to more effectively plan and allocate your client’s cross-platform video investment
via
The OMG Advanced TV Platform, powered by VideoAmp, a component of Omni
that
Leverages linear TV and digital data to plan, optimize and measure integrated video investments
and
Uncovers duplication of audiences to prompt more effective allocation decisions
in order to
Move beyond broad demos by using audience data to better plan video investments and fuel true cross-channel measurement against KPIs.