Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
©2017 Acxiom Company Confidential
MAY 2017
FROM FOCUS GROUP TO
DIGITAL REACHUSING ACXIOM DATA AND MODELLING IN R TO CREATE
CUSTOM FACEBOOK AUDIENCES
©2017 Acxiom Company Confidential
• Picture of Facebook adds
2
©2017 Acxiom Company Confidential 3
©2017 Acxiom Company Confidential 4
©2017 Acxiom Company Confidential 5
©2017 Acxiom Company Confidential
ACXIOM CONNECTS OFFLINE AND
ONLINE
6
InfoBase®
©2017 Acxiom Company Confidential
ACXIOM CONNECTS OFFLINE AND
ONLINE
7
InfoBase®
©2017 Acxiom Company Confidential
National File Covering
over 90% UK Households
InfoBase® Lifestyle Universe
55% of data is “actual” sourced directly from the consumer. Sophisticated models impute this out across the UK so all attributes have full coverage
Unparalled UK Consumer Base - designed to drive highly targeted customer acquisition and segmentation
• 90% UK coverage ensures optimum enhancement rates and marketing scale
• Highly accurate, market relevant predictors against every individual ensuring solutions can be deployed consistently across the entire base
• More actual data sourced directly from the consumer than any other national file
• Ensuring that data is predictive within solutions and able to deliver actionable insight
9 ©2016 Acxiom Company Confidential
˜ GenderMarital Status
˜ Year of birth˜ Full date of birth
Age (banded)˜ Partner's Year of birth
˜ Partner's Full date of birthIndividual's Lifestage - Age driven
˜ Household Level Lifestage - Age drivenIndividual's Lifestage - Family status driven
˜ Household Lifestage- Family status driven˜ Young Adult still living at home
˜ Number of Young Adults still living at home Parent Status
Dependent Children in Household
Number of children at home (0-21 years)
Number of children in the household aged 00-10
Number of children in the household aged 00-16
Number of children in the household aged 11-16
Number of children in the household aged 17-21
Child at home 0-4 years old
Child at home 5-7 years old
Child at home 8-10 years old
Child at home 11-16 years old
Child at home 17-21 years old
TV Region
County
Region Code
Country Code
Population Density
Combined annual household income
Equivalised Income
Equivalised household Income indexed to UK
Net household income per week (banded)
Net household income per week indexed to UK
Discretionary household income per week
Discretionary household income indexed to UK
Dual income no kids yet
Affluence Ranking
Household Affluence Ranking
Lifestage by Affluence
Household Level Lifestage by Affluence
9
Gender
Marital Status
Year of birth
Full date of birth
Age (banded)
Partner's Year of birth
Partner's Full date of birth
Individual's Lifestage - Age driven
Household Level Lifestage - Age driven
Individual's Lifestage - Family status driven
Household Lifestage- Family status driven
Young Adult still living at home
Individual's Occupation
Partner's Occupation
Individual's Employment Status
Partner's Employment Status
Self Employed
Partner is Self Employed
Run Business From Home(you)
Run Business From Home(Partner)
Number of students in household
Household Socio Economic Classification
Incomes across individual and partner
Number of earners in the household
Proportion of adults earning
Number of unemployed in the household
Proportion Adults unemployed
Number of non earning adults in household
Household Employment status
Pensioner Status
Motorist
Bought a car under 3 years old
SMMT Car Classification
Age of car
Bought car new/used
Number of Cars in Household
Car Fuel Type (petrol/diesel)
Annual Mileage
Car Make
Car Model
Car Year/Month Reg
Car Insurance Expiry Month
Years No Claims Discount
Buildings Insurance Expiry Month
Contents Insurance Expiry Month
Changed home insurance provider in last 3 years
Level of motor noclaims discount
Have a mortgage
Individual/Partner has Personal Loan
Individual/Partner is Credit Card Holder
Household Credit Card Ownership
Number of Credit Cards
Have Visa/Master card
Have American Express card
Have a Store/Shop Card
Have a Debit Card
Private Pension
Regular Savings Plan
Child Savings Plan
Unit Trusts/High Interest Investments
Own stocks/shares
Have an ISA
Investment Activity Ranking
Household level Investment Activity Ranking
Life Assurance
Private Medical Insurance
Accident Insurance
Funeral Plan
Insurance Activity Ranking
Household level Insurance Activity Ranking
Charities/Voluntary Work
Charity Donor Ranking
Household level Charity Donor Ranking
Donate to Environmental/Animal/Wildlife Causes
Donate to Global Causes
Donate to Local Causes
Contribute to charity in the Street/at the Door
Contribute to charity by Post
Quality newspaper Readers
Mid Market newspaper Readers
Popular newspaper Readers
Daily Record
Daily Express
Guardian
Independent
Daily Mail
Daily Mirror
Star
Sun
Daily Telegraph
Times
Financial Times
Take UK Holidays
Take European Holidays
Take USA Holidays
Take Rest of the World Holidays
Foreign Travel as a regular hobby
Snow Skiing as a regular hobby
Head of Household Indicator
Household Size-number of adults in household
Total Household Size (adults and children)
Summary Household Composition
Detailed Household Composition
Number of adults at home during weekdays
Green status – ranked percentile
Main Shopping Weekly Grocery Spend
Store Used For Main Shopping
Stores Used For Grocery Shopping
Shopping by Catalogue Interest
Mailorder frequency
Home Ownership Status
How many times homebuyer
Year moved to Address
Length Of Residence (Banded)
Month Moved Into Current Home
Year current household moved in
Household Length of Residence
Type of Property
Number of Bedrooms
Date Home Built
Have a PC in the Household
Have internet access at home
Have internet Broadband
Personal Computing as a regular interest
Games Console
Digital Camera
iPod/MP3 Player
Have flat screen TV
Have HD TV
Pay to view TV subscription
Cable TV
Satellite TV
Mobile Phone
Mobile Contract Payment Type (contract/pre-pay)
Household Technology Ranking
Consumer Electronics Audience Segmentation
Telecoms Audience Segmentation
InfoBase® Variables
Probability to buy groceries online - often
Probability to buy groceries online - sometimes
Probability to buy groceries online - never
Probability to buy insurance - online
Probability to buy insurance - in shop
Probability to buy insurance - by phone
Probability to use internet for - email
Probability to use internet for - Google
Probability to use internet for - eBay
Probability to use internet for - weather news
Probability to use internet for - price comparison
Probability to use internet for -social networking
Probability to use internet for -messaging
Probability to use internet for - gambling/betting
Probability to use internet for - games playing
Probability to use internet for - paying bills
Probability to research tech prod - online
Probability to research tech prod - in shop
Probability to research tech prod -from catalogue
Probability to buy technical prod - online
Probability to buy technical prod - in shop
Probability to buy technical prod - via catalogue
Probability to use mobile phone for - internet
Probability to have Freeview TV
Probability to have satellite TV
Probability to have cable TV
Probability to read news online - often
Probability to read news online - never
Probability to read news online - sometimes
Probability to book holiday via - Internet
Probability to book holiday via - Agent
Online purchase frequency
Bet on Horse Racing
Book Reading
Charities/Voluntary Work
Crossword Puzzles
Current Affairs
Cycling
Do-It-Yourself
Eating Out
Fashion Clothing
Fine Art/Antiques
Football
Foreign Travel
Further Education
Gardening
Going to the Gym
Going to the Pub
Golf
Cooking/Fine Foods & Wines
Grandchildren
Health Foods
Hiking/Walking
Household Pets
Jogging/Physical Exercise
Music/CD's
National Trust
Personal Computing
Prize Draws & Competitions
Religious Activities
Snow Skiing
Theatre, Cultural/Arts Events
Vitamins/Food Supplements
Wildlife/Environmental Concerns
Non Smoking Household
Cultural Pursuits Interest level
Entertainment Interest level
Animal/Nature Awareness level
Outdoor Pursuits level Prefer marketing communications via Email
Prefer marketing communications via Direct Mail
Prefer marketing communications via Telephone
Prefer marketing communications via Text
Prefer marketing communications via Social Networking
Channel preference mix segments
InfoBase® Affordability
Personicx
©2017 Acxiom Company Confidential
ACXIOM CONNECTS OFFLINE AND
ONLINE
10
InfoBase®
© 2016 LiveRamp. All rights reserved.
OFFLINE IDENTITY DIGITAL IDENTITY
De-identification
via secure
one-way hash
R. Jones
10 Main St
Plano, TXRebecca Smith
IP 222.58.1.10
Becky
Wilson
20 Stag Dr
Chicago,
IL
Becky Smith
I23 Sunrise Ave
Plano, TX
Phone 555-1212
Rebecca Smith
Cell phone: 555-
2323
Rebecca Wilson
20 Longhorn Ave
ABILITEC
LINK
abc12345
IDENTITY
LINK
a7b9j7ag
LiveRamp IdentityLink
Identity is the Key to People-Based Marketing
How Does IdentityLink Work?
1212
Data Sources Destinations
TRANSACTION
DATA
CRM
DATA
3rd PARTY
DATA
DATA
MANAGEMENT
MEASUREMENT
&
ATTRIBUTION
IdentityLink
IDENTITY RESOLUTION
RESOLUTION & ONBOARDING
DATA STEWARDSHIP &
SECURITY
OTHER 1st Part2nd PARTY
DATA
MEDIA PLATFORMS
SEARCH
VIDEO
MOBILE
EXPOSURE
DATA
© 2016 LiveRamp. All rights reserved.
MOBILE
MOBILE
MEASUREMENT
DATA
MANAGEMENT
VIDEO
DATA
MANNAGEME
NT
SEARCH
CRM &
DATA
PROVIDERS
SITE OPTIMIZATION
MEDIA
PLATFORMS
ADDRESSABLE
TV
1:1 MARKETING &
IdentityLink
resolves identities across channels,
devices, and over 400 partner destinations
©2017 Acxiom Company Confidential
ACXIOM CONNECTS OFFLINE AND
ONLINE
14
InfoBase®
©2017 Acxiom Company Confidential
1) FIND THE PANEL PARTICIPANTS IN
INFOBASE
15
©2017 Acxiom Company Confidential
2) FEATURE SELECTION
16
Gender
Guardian Reader
Affluence
Age
Teenage children
Lives in city suburbs
©2017 Acxiom Company Confidential
3) CREATE A CLASSIFICATION MODEL:
PREDICT AFFINITY USING INFOBASE
VARIABLES
17
©2017 Acxiom Company Confidential
4) USE MODEL TO SCORE UK
POPULATION
18
©2017 Acxiom Company Confidential 19
Audience
filtering
Feature
selection
Model
buildingScoring
Manual
Filtering
Manual IV and
correlation
checks
Stepwise
Logistic
regression
Serial (3 hours
run time)
Automatic
filtering Regularised regression (lasso)
Distributed
over Hadoop
One of the first tasks of the data science
team was to move this process from SAS to
R
©2017 Acxiom Company Confidential
Gains are releasing people and time
20
2
days
20
Minutes
©2017 Acxiom Company Confidential
Self service is our next goal
21
Automate
filtering
Distribute
scoring
Create UI
(prototype
in shiny)
©2017 Acxiom Company Confidential
Questions?
©2017 Acxiom Company Confidential
We are hiring!
Data Scientists
Data Engineers
Insight Analysts
23
©2017 Acxiom Company Confidential