Module 3 Adapative Customer Experience Final

Real-Time Analytics &
Attribution

• Noah Powers
– Principal Solutions Architect, Customer Intelligence, SAS

• Patty Hager
– Analytics Manager, Content/Communication/Entertainment, SAS

• Suneel Grover
– Solutions Architect, Integrated Marketing Analytics & Visualization, SAS
– Adjunct Professor, Business Analytics & Data Visualization,
New York University (NYU)

The Digital Self

http://youtu.be/N17Ck6W73c8?hd=1

Module 3
Adaptive Customer
Experience

Adaptive Customer Experience

Marketing
Decisions
Customer Experience Analytics
Customer Experience Targeting & Personalization
Social Media Analytics
Case Studies

Information Management & Analytics

ERP CRM EDW Online Social Other

Data Sources

Customer Experience Management

“A solution that enables the management and delivery of dynamic,
targeted, consistent content, offers, products, and interactions
across digitally enabled consumer touch points."

“The Emergence Of Customer Experience Management Solutions”
Delivering Cross-Touchpoint Customer Experiences Drives Need For New Capability

© 2011, Forrester Research, Inc. Reproduction Prohibited

CXM-Driven Sites Represent A Very Different Site Management Paradigm


CXM Solutions Will Emerge From The Convergence Of Many Solution Categories


“What Does The Web Analytics Industry Want To Be When It Grows Up?”
Web Analytics Supports Marketing


The Challenge Of Traditional Web Analytics
Enterprise digital measurement has been driven by
two major trends:
1. The use of tags to collect user behavior
 Time consuming, difficult, and cumbersome
 Not a typical IT activity

1. The reliance of SaaS vendors to provide aggregated
reporting for digital marketing
 Lack of deep data access
 Little or no ability to perform advanced analytics

“Both of these trends are
at crisis point in 2012.”
12

Copyright © 2011, SAS Institute Inc. All rights reserved.

What We Hear From Clients
 Partner: Adobe Omniture, WebTrends, etc.
 No real-time data options
 1-2 day data access delay

 Web Behavioral Data for ARPU & Churn
 Current State: Page view level
» “Data structure is challenging…not traditional DBMS”
» “Not receiving level of detail appropriate for predictive
modeling”
» “Proactive marketing limited by web data availability”

 Constraints

13


Problems Create Opportunity

Opportunity #1 Opportunity #2 Opportunity #3

Ability To Collect Pre-Processing Data Leveraging Adv. Analytics
& Own Granular Data For Marketing & Analytics & Marketing Automation

14


Business Challenge Themes

1. How can we improve our time-to-readiness?
2. How can we reduce the amount of time needed to
stitch together online & offline data?

Opportunity #1
 Web Behavior Table: ~ 1 day lag
 Data Warehouse Load: ½ -1 day lag

Ability To Collect
& Own Granular Data
15


Web Behavior Daily Table Extracts

Opportunity #2

 After ~2 days has passed…
 Data is still not ready
1. Raw state
2. Need to roll up to
session, visitor and
Pre-Processing Data subscriber level
For Analytics & Marketing
16


Remember - Where We Want To Get To…

CRM Data Enrichment Data
Integrated Marketing
Data Table
(Customer ID, 12345)
(Name, John Smith)
(Gender, M)
(Age, 42)
(Life Stage, FL)
(HH Income, 75K-100K)
(Children Ind, 1)
(HH Education, College)
(HH Value Score, Above Avg)
(CC Propensity, 0.57)
(Visit Recency, 12)
(Session Count, 7)
(Session Avg. PV, 4)
(Engagement, High)
(Content Goal, 1)
(Sticky Goal, 0)
Online History Data (Session Affiliate, Org Search) Current Session Data

17


Digital Data
Collection Process

18


Dynamic Data Collection
 Global Tagging: Collection at the browser level by
adding one insert, that never changes, to each page of
your site:
<SCRIPT language="JavaScript"type="text/javascript“src=“…………….."></SCRIPT>

 Dynamic content recognition
 Automatic collection of ALL activity (Subscriber or Anonymous)
 Highly accurate granular data - timed to the millisecond

 Dynamic collection rules
 Choose to collect the level of detail to meet your PII Policy

19


Real-Time Data Capture

20


Stitching Web Sessions Together

Customer (name, email, account id, etc.)

Visitor (Digital ID) Visitor (Digital ID)

Session 1 Session 2 Session 1
3 Session 2
4 Session 3
5

21


Digital Data
Transformation Process

From Data to Decisions:

~80% Data ~20% Analysis
Access/Preparation

22


Transform: Data Pre-Processing

 Pre-built data preparation processes (i.e. automation)
 Configurable rules which can be operationalized,
monitored, and adapted over time
 Data integration with online (real-time or batch) and offline
data streams focused on the individual
» Log in info and/or PII data defines match key

23


Transform: What’s Possible?
Goals Achievement Campaigns Promotion Analysis

Understand what is driving Improve understanding of Understand the effectiveness of
visitors to achieve specific goals marketing attribution and promotional / merchandising
during their visit effectiveness content within digital property

Transaction Analysis User Type Analysis Form Analysis

Understand multi-step Understand how people arrive Which forms are used, problem
processes including drop out and see the value of what they fields, entered data, time to
rates, completion rates, errors, do through behavioural scoring complete, and abandonment
etc. rates

Content Analysis In-Site Search

Which areas attract what types How visitors use in-site search?
of visitors? Where they leave How successful they are finding
from? How long do they stay? content, products, and services?

24


Transform: Batch

25


Transform: Real-Time

26


Post Transformation

Customer

Visitor
(Persistent Cookie)

Session
(Session Cookie)

27


Advanced Analytics

28


What’s At Stake?
ANALYTICS
TEXT ANALYTICS

Finding treasures in unstructured data
FORECASTING like social media or survey tools
that could uncover insights
about consumer sentiment
Leveraging historical data
to drive better insight into
decision-making
for the future

BUSINESS
INTELLIGENCE
OPTIMIZATION

Analyze massive
DATA MINING
amounts of data in
order to accurately
Mine transaction databases
identify areas likely to
for data showcasing patterns
produce the most
beneficial to marketing
profitable results

STATISTICS 29


Digitally Enhanced Analytics
Decision Trees Survival Analysis
(Bootstrap Forests/Boosting) (Retention / Churn)

Two-Step Models

Clustering Association / Sequence
(Proprietary Segmentation) (Acquisition / Cross-sell)

30


Remember - This Is What You Want

Probability scores are the output of predictive
models, and are an essential ingredient to
making data driven decisions


Let’s Imagine A Real-Time
In-Session Use Case….

32


DMA Prospect: Acquisition

Explores Registration & Returns Back To
Prospect Visits Website
Partially Completes Home Page

? © 2011, Forrester Research, Inc. Reproduction Prohibited

Application For Real-Time Prospecting
Prospect Returns To
Explores Registration
Home Page

On Line Behavior

Personalize Website
Personalize External Create Real-Time Trigger Advanced Analytics &
Experience
Experience (Re-Marketing) (In-session) Business Rules

DMA
Register & 2012
Save $500 Register
Now


Imagine The Opportunity For Relevant & Creative
Execution That The DMA Could Leverage With
Returning Conference Attendees…


Real-Time Triggers & Content Personalization

Register &
Save $500
Trigger Rule


The Interactive Advertising Bureau Perspective

http://youtu.be/bqgjJ-2dGeo?hd=1
37


Social Media, Analytics, & Visualization

http://youtu.be/0qt3WpAofI8?hd=1
(Stop Time: 6:32 Time Stamp)

Different Views Of The Social World
Consumer Mindset
 The need to be connected and the internet as a way
of life

 Corporate Mistrust and need for alternate advisors

 Power of the consumer voice and social media as the
global soapbox

Business Forces
 Impact of corporate reputation on business health

 Increased public scrutiny of business

 Greater fight for loyalty and advocacy

 Speed of information and brand impacts

Influence On Consumer Buying Process Is
Stronger Than Ever

• 1.5B+ people online and 1B+ using
Social Media Is social media monthly around globe.
Impacting Your Brand
• Consumers relying on social media and
connections to shape buying decisions.

• Public evaluation of brands is impacting
business health.

• The real brand sentiment is “out there”
in blogs and commentary and that data is
doubling every 18 months.

• And the answers and implications of that
for your business are in the data…

It’s Important To Engage Customers Where They Are

• Social Media is
everywhere – it’s not
just Facebook and
Twitter.

• Your customers are
there talking about Social Media
your brand. Conversations

• What are customers
saying about you and
what impact could
that have on your
business?
Source: The Conversation: Brian Solis and Jess3

Attitudes Toward Social Media
The use of social media by our organization 69%
will grow significantly

Our organization has a significant learning 61%
curve to overcome

Interest in utilizing social media is growing 57%
rapidly within our organization

Until we can measure social media, it will not be 50%
taken seriously

Social media is an important component of 46%
our overall marketing strategy

Use of social media by our organization is a 45%
tactical rather than strategic decision

Using social media is integral to our overall 42%
company goals and strategy

Social media has been designated high priority by 32%
our organization’s executives

It is difficult to see the value of social 29%
media for business purposes

Social media tools are not very relevant for our 21%
business
The use of social media for business 11%
purposes is a passing fad

Worlds Are Already Colliding….

Search Engine Marketing
Website Design/Experience
Website E-Commerce
Online Campaign Mgmt
Analytics

Media Social
TV Mix Media
Radio Topic Analysis
Print Sentiment Analysis
Online Social Volume Analysis
Out of Home Surveys & Voice of Customer

Social Media Analytics Maturity Model

What ROI are we getting on digital campaigns? Have we lowered risk? Social
Are we achieving financial and brand-related goals? Scorecard
Competitive Advantage

Based on trends, where are conversations going?
How can our plan optimize impact on traffic, sales? Planning &
Who are the right influencers? Engagement

Historic How do today’s online mentions compare to
Analysis previous period? … same period 2 years ago?

Attributed What are people saying about our customer service?
Listening …our product quality? … our prices? …our selection?

Basic
What are people saying about us? About our competition?
Listening

Degree of Intelligence

Business Challenges In Social Media
• WHAT are consumers saying
about your brand? About the
• WHERE are consumers talking? competition?
• Is volume trending up or down?
• WHAT aspects of your business
• WHICH sites matter most?
drive satisfaction and loyalty?
• WHICH sites are more positive? • WHAT questions and unmet
Negative?
needs emerge?

• WHO is creating content
about your brand…Journalists?
• HOW do perceptions differ Bloggers? Forum members?
across the various channels
through which customers give you • WHO among these authors is
feedback? a threat to reputation? An
opportunity for advocacy?

Social Media Analytics Process Flow

Listen Engage Leverage
• Understand customer • Establish relationship • Deliver value back to
sentiment, trends, with “community” and consumers – show that
issues and consumer, and gain you heard them
opportunities their trust
• Leverage insight and
• Unearth emerging • Introduce your brand feedback and port back
topics, phrases or voice into the across all aspects of your
issues relevant to conversation in an business from brand, PR,
your business authentic way customer service to
marketing

What Matters For Social Media Excellence
 Data Coverage
 Establish direct link to multiple data aggregators and proprietary
social media channels… and online channels
 Data Quality & Relevance
 Higher quality data (manage the signal-to-noise ratio)
 Relevance of data to business issues
 Feeds into marketing measurement framework (media mix)
 Continuously improving accuracy of data processing
 Sentiment analysis, concept extraction, entity identification
 Iterate quickly on preliminary results then refine rules
 Deliver results… deliver ANSWERS
 Map insights across departments, enterprise

What Matters For Social Media Excellence (2)
 Focus on quality of data / results Awareness

 Relevant conversations
Consideration
» Exclude irrelevant mentions
 Filtering out noise
Intent
 Focus on action
 Consumer insights Web Traffic

 Market insights
Store
 Media insights Traffic

 Engagement across social channels
Sales

Relevant Social Data
Blogs Social Platforms
• 2M+ blog posts and  Multiple years of archived
comments
online conversation history
(where available)
 Incorporate new
websites/data sources as
Message Forums Online News they emerge
• 5M unique forum URLs • Coverage of >40,000
traditional media  Global organizations need
sources support for multiple
languages
 Support both external (blogs,
Review Sites Customer Care message boards, reviews,
& VOC Data news) and internal text data
• >100k reviews and
posts per day
• Call logs sources (company
• Customer Emails
• Web chats forums/blogs, survey open-
• Survey open-ends ends, call center logs, etc.)
• Front-line feedback

Taxonomy/Rule Engines

Industry Base
Retail Bank

Online Retail
Language Base
Telco English

Hospitality
Spanish

French
International Base
Source- Anti-Spam
German Emoticons driven
Places (city,
( ) “Bad Words” state, Twitter & Txt
Automotive vernacular
(open country, Speak
Japanese Symbols (!!) (like Twitter-
region)
speak) sourced)
….
….

Classification / Taxonomy Example (Telco)
Customer Long hold time; friendly; helpful; rude;
Service knowledgeable; impatient

Value & great bargain; good deal; lots of minutes for less;
Pricing costs too much; cannot believe my bills; too many
fees; roaming charges; bundles; triple plays

Network spotty coverage; no signal; I get all bars; great sound;
Mobile calls dropped
Coverage

Network TXT messaging; GCASH; Mobile browsing; roaming;
Services content downloads; ring tones; nice screen; easy to
use; looks great; too confusing; buttons too small

Personal Phone nice screen; easy to use; looks great; too confusing;
Line of Business Features buttons too small

Customer Long hold time; friendly; helpful; rude;
Service knowledgeable; impatient; knowledgeable technician

Television & Value & great bargain; good deal; costs too much; cannot
Broadband Pricing believe my bills; too many fees; nickel and diming for
every feature; bundles; triple plays

Network Installation process; satellite; cable; IP TV; hazy
Services picture; clear resolution

Sentiment Analysis / Classification Approach
 Taxonomies and Sentiment Analyses should be
customized for each customer to ensure best
alignment and accuracy possible
 Is critical to ensure that the raw data collected is
cleaned and organized in order to effectively derive
insights
 Important to measure both document-level and
attribute-level sentiment using a hybrid of two
primary methods:

Leverage advanced
text analytics
 Statistical Method (Bayes Algorithm, BM25, Pivoted Length
Normalization, Smoothed Relative Frequency and Relative
Frequency)

Leverage human  Rule-Based Method (wide range of Boolean operators used to
intelligence & experience develop manual rules for additional control and customization)

Benefits To Social Engagement

Enable customer care agents monitoring social media to
communicate with consumers in order to…

Address the Mitigate or Reinforce a Facilitate
consumer’s respond to customer’s customer
service negative positive consideration
issues and comments sentiment process
questions or threats

Be Part Of The Conversation
Enable your organization to participate in online
conversations by leveraging analytical insights.

1. Presents social media messages that are relevant
2. Prioritize and dispatch messages to appropriate
team members
3. View customer information and messages to get
a complete context of the conversation
4. Collect data regarding the conversation to monitor
effectiveness of the interaction
5. Apply appropriate treatment to customer concern

Respond To Authors “In Channel”
Active feedback channel for events, launches, promotions

This project investigates how social media and online user-generated content
can be used to enrich the understanding of the changing job conditions in the
US and Ireland by analyzing the moods and topics present in unemployment-
related conversations from the open social web and relating them to official
unemployment statistics.

Two specific questions were addressed:

1. Can online conversations provide an early indicator of impending job
losses?
2. Can online conversations help policy makers enrich their understanding of
the type and sequence of coping strategies employed by individuals?

Module 3 Adapative Customer Experience Final

Sunday Afternoon Preview
• Understand how to align activities and resources to
strategies and goals
• Know how to establish accountability by linking
marketing performance to financial measures
• Recognize a sound approach to optimizing the
marketing mix

Strategy & Planning

Marketing
Decisions
Marketing Mix Optimization
Marketing Operations Management
Case Studies

Information Management

ERP CRM EDW Online Social Other

Data Sources

Module 3 Adapative Customer Experience Final

More Related Content

What's hot (20)

Viewers also liked (15)

Similar to Module 3 Adapative Customer Experience Final (20)

More from Vivastream (20)

Module 3 Adapative Customer Experience Final

Editor's Notes