Making Big Data Smart

Copyright©2017TheNielsenCompany.Confidentialand
proprietary.
proprietary.
NIELSEN
Making Big Data Smart
PROPRIETARY & CONFIDENTIAL
N
S
EW

proprietary.

proprietary.
3
Topics of Conversation
The Nielsen Panel
Big Data – What it isn’t
Nielsen Common Homes
Ensuring Quality Measurement

proprietary.
Nielsen National People Meter Homes

proprietary.
Nielsen National People Meter
• 40,000+ homes, 100,000+
persons
• Homes can be in the panel up to 2
years
• Electronically measures both what
the set is tuned to and who is
watching
• Each household member assigned
a button (additional buttons
available for visitors)

proprietary.
Nielsen Meters
Nielsen’s Global Television Audience Meter
(GTAM)
• Next generation metering platform
• Meets the challenges of digital media
• Adapts to a varied home viewing environment
Scrolling Text People Meter (STPM)
• Allows for dynamic prompts
• Supports multiple languages
GTAM
GTAM Lite
Code Reader
Scrolling Text
People Meter

proprietary.
7
WHAT BIG DATA ISN’T

proprietary.
Measurement relies on COMPLETE MARKET COVERAGE, which means
all geographies, all population segments & all types of viewing environments
Nielsen Panel
DESIGNED FOR MEASUREMENT
Big Data
NOT A CENSUS
Big Data Isn’t a Census of the Market

proprietary.
NOT ALL TVs HAVE SET-TOP
BOXES
NOT ALL SET-TOP BOXES ARE
RETURN PATH CAPABLE
DOES NOT CAPTURE ALL
ACTIVITY AT THE VIEWING SITE
THE SCOPE OF AVAILABLE DATA
VARIES BY PROVIDER
SET TOP BOX DATA IS NOT “PLUG
AND PLAY” INTO TELEVISION
RATINGS
SET TOP BOX DATA
Tuning data that cable and
satellite providers can collect
from their subscribers’ set-
top boxes
Big Data Isn’t a Census of Itself

proprietary.
STB HUT Rating LPM HUT Rating
STB IS ON
TV IS ON
SPIKES ARE PRIME TIME
Panels are essential to model on/off across all the different RPD boxes
STB Data Isn’t Necessarily Viewing

proprietary.
Big Data Home
GENDER
EDUCATION INCOME
PRESENCE
OF CHILDREN
LANGUAGE
SPOKEN
OCCUPATION
ETHNICITY
# TV SETS # SMARTPHONES# TABLETS
AGE
PHONE
/INTERNET
PROVIDER
VISITORS
GAME CONSOLES SVOD VOD
MARITAL
STATUS
Big Data Doesn’t Have Demographics
% CLASSIFIED CORRECTLY BY THIRD PARTY
HOUSEHOLDS WITH…
Persons Age 18-34
Persons Age 35-54
Hispanic
HH Size 1
HH Size 2
HH Size 4+57.7%
70.6%
59.9%
14.1%
28.4%
49.9%
27% OF THIRD-PARTY MATCHED
HOUSEHOLDS ARE MISSING KEY
CHARACTERISTICS

opyright©2016TheNielsenCompany.Confidentialandproprietary.
12
Big Data Doesn’t Know Who’s Viewing?
Fiel
d
Rep
GAME
CONSOLE
ROKU

proprietary.
proprietary.
CORRECTED INPUT DATA
Source: Nielsen Common Homes, May 2016
Includes all Reportable Networks and Stations for the May 2016 NSI Measurement Period
Network and Station Crediting – Time-Shifted Viewing
74% ENTITIES CORRECTLY IDENTIFIED BEFORE NIELSEN FIXES
ENTITIES CORRECTLY IDENTIFIED AFTER NIELSEN FIXES 96%
ONLY HIGH-QUALITY PANELS CAN UNCOVER THESE ERRORS

proprietary.
14
NIELSEN COMMON HOMES
DETERMINE THE ACCURACY OF BIG DATA SOURCES
SIDE-BY-SIDE COMPARISON: PANEL SERVES AS TRUTH SET
CORRECT INACCURACIES: IN HOUSE OR VIA DATA PROVIDER
PERFORM PANEL MATCH FOR EACH PROVIDER, ON ONGOING BASIS

proprietary.
15
NIELSEN PANEL HOME

proprietary.
16
NIELSEN PANEL HOME
SET TOP BOX DATA
NIELSEN METER DATA
SAME HOME/SAME TVs

proprietary.
17
NIELSEN PANEL HOME
SET TOP BOX DATA
NIELSEN METER DATA
X
SIDE-BY-SIDE COMP

proprietary.
proprietary.
COMPLETE HOUSEHOLD CHARACTERISTICS
Source: Nielsen Panel Homes, May 2016
Enhancing and Correcting Third Party Data
THIRD PARTY DATA ALLOWS US TO ASSIGN HOUSEHOLD CHARACTERISTICS…BUT IT ISN’T
PERFECT
? ? ?
27% OF THIRD-PARTY
MATCHED HOUSEHOLDS
ARE MISSING KEY
CHARACTERISTICS

proprietary.
proprietary.
FULL MARKET REPRESENTATION
RATINGS BASED ON RETURN-PATH DATA ONLY ARE INCOMPLETE
NOT ALL SET-TOP-
BOXES RETURN
DATA
RETURNED DATA ALONE DOES NOT
COVER ALL AUDIENCES
WHY DOES THIS MATTER?
THERE ARE IMPORTANT DIFFERENCES BETWEEN HOMES THAT
RETURN DATA AND THOSE THAT DO NOT

proprietary.
proprietary.
19%
MORE LIKELY TO
EARN $75,000+
10%
LESS LIKELY TO BE
HISPANIC
18%
MORE LIKELY TO
HAVE A COLLEGE
DEGREE
RPD-CAPABLE HOMES ARE
RATINGS BASED ON UNADJUSTED RETURN-PATH DATA
DO NOT ACCOUNT FOR ALL AUDIENCES

proprietary.
proprietary.
Implications for Ratings
1.1 1.9
1.2 0.3
1.3 1.2
NON-RPD HOMES RPD HOMES
AFFILIATE A
AFFILIATE B
AFFILIATE C
RPD-ONLY RATINGS ARE
INCOMPLETE
A HIGH-QUALITY PANEL ALLOWS
NIELSEN TO ADJUST TO THE TOTAL
MARKET

proprietary.
proprietary.
ENSURING QUALITY MEASUREMENT
END-TO-END COMMITMENT TO QUALITY
BIG DATA PLUS A HIGH-QUALITY, GOLD-STANDARD PANEL
RESULTS IN RETURN-PATH DATA DONE RIGHT
CORRECT & COMPLETE
HOUSEHOLD
CHARACTERISTICS
CORRECTED INPUT
DATA
PERSONS-LEVEL
ESTIMATES
FULL MARKET
COVERAGE

Making Big Data Smart

More Related Content

More from Domino Data Lab (20)

Recently uploaded (20)

Making Big Data Smart