Do's and Don'ts of Data Driven Marketing
Transcript of Do's and Don'ts of Data Driven Marketing
1
Mike Gualtieri Principal Analyst Serving Application Development & Delivery Professionals, Forrester Research Steve Dille SVP and Chief Marketing Officer, Message Systems
Do’s and Don’ts of Data Driven Marketing
2
A link to the webinar replay will be provided via email following the presentation
2
Participate in Today’s Discussion… Tweet #datainsights
Follow us on Twitter @messagesystems @Forrester
Follow us on Linkedin Message Systems Forrester
Big Data Trends And The Future Of Digital Marketing Mike Gualtieri, Principal Analyst November 18, 2014
© 2014 Forrester Research, Inc. Reproduction Prohibited 4
Firms know that consumers want to be treated as individuals
3.8%
26.2%
44.7%
47.3%
50.6%
54.4%
55.3%
55.8%
58.4%
0% 10% 20% 30% 40% 50% 60% 70%
None of these
To interact with other customers
To serve themselves
To be heard by us (give us feedback)
Consistent treatment across channels
To deal with us via their smartphones
More information and help from us
Easier interactions with us
To enjoy their dealings with us more
To be treated as individuals
Base: 423 IT/marketing professionals, Q1 2013; Source: Forrester Research/IBM Customer-Facing Applications Survey Q1, 2013
“What new customer expectations does your organization face?” (Check all that apply for each category of customer)
#Royalty
Trend
Consumers want to be treated like royalty.
Royalty experiences have contextual awareness and adapt to serve a
single consumer.
Treat consumers like royalty to get their loyalty.
© 2013 Forrester Research, Inc. Reproduction Prohibited 11
How well do you know this consumer? › Male
› 35 years old
› Single
› Resides in New York City
› Makes $100,000 per year
What do you predict he would do if the bank accidently transferred $5,000 into his bank account?
A. Give the money back B. Take the money and run
Quiz
George Costanza
“The simple analysis of aggregated customer interactions
with digital channels is not enough for firms to truly achieve
personalized marketing”
One-to-one One-to-many One-to-some One-to-one
CRM
Big Data
Personal Relationships
Mass Production
Cus
tom
er E
xper
ienc
e
#BigData
Trend
Big Data means all your data++
© 2013 Forrester Research, Inc. Reproduction Prohibited 17
30%
7%
12%
21%
30%
9% 8%
14%
35% 34%
The term “big data” is very confusing; not sure what it
means
It’s a bunch of hype with little substance and few new
ideas
It’s about new technologies that allow us to handle more
data
It’s an extension of existing analytics and BI practices
suited for data that is larger or faster than we are used to
It’s a whole new way of thinking about the value in
data that requires new analytics and leverages some new technologies
Business Decision Makers Technology Decision Makers
Base: 452 North American technology decision-makers Respondents answering “don’t know” are not shown Source: Global Data and Analytics Survey, 2014
Base: 249 North American business decision-makers Respondents answering “don’t know” are not shown Source: Global Data and Analytics Survey, 2014
Most technology decision makers get it; 30% of business decision makers are confused.
© 2013 Forrester Research, Inc. Reproduction Prohibited 18
Big Data gushes from many sources…
• Data described by a schema • Clicks, relational database, XML, delimited flat
file, system events
Structured text
• Free-form text • Email, documents, tweets, blog comments,
Facebook status, genome
Unstructured text
• Audio, images, video • Surveillance cameras, geological survey maps,
Siri voice Binary
© 2013 Forrester Research, Inc. Reproduction Prohibited 19
What percentage of enterprise data do firms use for analytics?
A. 12%
B. 34%
C. 53%
D. 76%
Enterprise Data
Quiz
© 2013 Forrester Research, Inc. Reproduction Prohibited 20
What percentage of enterprise data do firms use for analytics?
A. 12%
B. 34%
C. 53%
D. 76%
Enterprise Data
Quiz
Source: Forrester Research
1100
1001
1011
001
0100
1001
1011
001
0100
1100
1101
101
0100
1001
1011
001
Historical Transactions
Customer data
Ops
© 2013 Forrester Research, Inc. Reproduction Prohibited 22
A multi-dimensional view of the customer provides individualized attributes, behavior, and preferences.
#Predictive
Trend
Predictive models can predict customer attributes and behavior.
ANALYTICS
PREDICTIVE Techniques, tools, and technologies that use
data to find models – models that can anticipate outcomes with a significant
probability of accuracy.
Predictive models can have a delightful, multiplicative effect on the bottom line
Direct Marketing – 1% response rate Send marketing mail to 1,000,000 customers at cost of $2 per mailing to sell a $220 service.
$2 x 1,000,000 $2,000,000
1% response rate means 10,000 customer will buy service
$220 x 10,000 $2,200,000
Profit* $200,000 Predictive Direct Marketing – 3% response rate Send marketing mail to 250,000 customers predicted most likely to buy at cost of $2 per mailing to sell a $220 service
$2 x 250,000 $500,000
3% response rate means 7,500 customer will buy service
$220 x 7,500 $1,650,000
Profit when using a predictive model* $1,150,000
Traditional
Predictive
* Profit calculation does not include other expenses
© 2013 Forrester Research, Inc. Reproduction Prohibited 27
Predictive models can be very powerful and profitable, but understand that:
› Predictive models are about probabilities, not absolutes • E.g. 78% chance you will like Breaking Bad
› Predictive models may not exist for every question • E.g. Economists, elections, etc…
But, when they work they give your firm an “unfair” advantage.
© 2013 Forrester Research, Inc. Reproduction Prohibited 28
Data scientists use a combination of statistical
and machine learning algorithms to find
patterns and predictive models.
What customers are thinking about switching carriers?
Churn
© 2013 Forrester Research, Inc. Reproduction Prohibited 30
Predictive analytics is very different from traditional analytics
Traditional Analytics Predictive Analytics
• Choose a business outcome to improve
• Discuss and decide what data will be relevant
• Develop a data model
• Design reports and dashboards
• Choose business outcome to improve
• Assemble all possible data
• Run algorithms to find relevant data & predictive
model
• Use the predictive model
How can you provide a perfect , individualized product recommendation?
Recommend
Pitch
What descriptive text attracts buyers, creates urgency and supports the highest price?
Geo
How can I profit from the differences in regional shopping behavior?
Talk
What are people saying about products?
How can you listen to customers to measure share of voice, efficacy, and side effects?
Sentiment
Turkey
How can you predict where customers will be on Thanksgiving.
Influence
Can you predict what voters will be positively impacted by campaign (marketing) contact?
#Streaming
Trend
Streaming customer data is flowing by, and value is slipping away.
Now
Should you make an offer she can’t refuse or send a gentle nudge?
How can Spotify use accelerometer data generated by customers while they listen?
Activity
#DigitalCreepiness
DEFINITION
FORRESTER A feeling by a customer that a digital experience offered by a
company knows more about them than they should and is using that information in a way that makes the customer feel
uncomfortable.
© 2013 Forrester Research, Inc. Reproduction Prohibited 44
There is a precarious balance between a personalized experience and digital creepiness.
“Google policy is to get right up to the
creepy line and not cross it”
-Eric Schmidt, Executive Chairman of Google
© 2013 Forrester Research, Inc. Reproduction Prohibited 46
Base: 5,012 US Online Adults (18+)
Some customers care, some don’t.
Only 33% of online consumers will stop shopping at an online retailer if they knew the retailer
was tracking their behavior.
© 2013 Forrester Research, Inc. Reproduction Prohibited 47
Customers can react to feelings of digital creepiness in different ways
#Challenges
Garbage In = Garbage Out
© 2013 Forrester Research, Inc. Reproduction Prohibited 50
Bi-directional digital delivery platform
› Data silos hinder visibility and prevent advanced insights.
› Big Data strains the ability of legacy technology to delivery personalized digital experiences
› Predictive and streaming analytics capabilities are lacking or non-existent
› There’s a lack of external and contextual data sources that enrich customer data.
#Opportunity?
Treat consumers like royalty to get their loyalty.
© 2013 Forrester Research, Inc. Reproduction Prohibited 53
They are ready.
© 2013 Forrester Research, Inc. Reproduction Prohibited 54
Now you must get ready by starting a technology revolution › Multi-dimensional customer data management
• To understand individual customers
› Predictive analytics regime • To predict individual customer behavior and attributes
› Streaming analytics • To capture contextual information in real-time
› Bi-directional digital delivery platform • To interact with customers and prospects
#Imagination
How personalized, how contextual, and how
real-time can you make your marketing?
Thank you Mike Gualtieri [email protected]
58
What customer communication channels are you currently using or do you intend on using in 2015 as part of your big data personalization strategy? (check all that apply)
A) Email
B) Display advertising
C) Website personalization or recommendations
E) None (we will still be gathering and building our big data)
59
CONFIDENTIAL
Steve Dille
Do’s and Dont’s of Implementing Big Data Driven Marketing
November 2014
60
Message Systems Business: E-Mail Infrastructure
WHO WE ARE Message Systems’ e-mail infrastructure software is the preferred choice of the world’s best known cloud businesses, e-mail service providers, telecom companies and large enterprises who depend on e-mail that is rich, relevant and real-time to drive revenue.
OUR PRESENCE Message Systems customers move over 2.5 trillion messages a year — more than 20 percent of the world’s legitimate email.
DIFFERENTIATION Flexibility, deliverability and control needed to drive the highest possible customer engagement and revenue. Unsurpassed performance and scalability.
61
Partial Customer List
Carriers Enterprise
Cloud SaaS Marketing Platforms
62
A Few Customers Driving Email With Big Data
63
A Big Transition is Happening
64
Old School: Campaign Oriented E-Mail Marketing
• Campaign Oriented • Episodic • Targeting: Segmentation, Batch Blast
ESP E-Mail CRM System
LIST
METRICS
BATCH
DATA SOURCES
CUSTOMERS
PROSPECTS
CHANNELS
65
• Real-Time, Continuous Engagement, Tight Data Integration • Targeting: Personalized 1X1, Behavior and Context Driven
New School: Big Data-Driven Marketing
CRM
E-COMMERCE HISTORY
SOCIAL INTERACTIONS
DEMOGRAPHICS
NODE
Hadoop Cluster
Predictive Analytics
BIG DATA CHANNELS
CUSTOMERS
PROSPECTS NODE
NODE NODE
NODE NODE
GPS Location
Website
Advertising
Mobile
Call Center
Content Marketing
In Store Marketing
CALL RECORDS
METRICS
CLICK STREAM
INTERACTIONS
POS DATA
METRICS
INTERACTIONS
66
Practical Lessons
67
Start with One Channel — Work in Pairs
NODE
Hadoop Cluster
NODE
NODE NODE
NODE NODE • Relevant
Targeting Data
• Analytics
CHANNELS
Website
Website
Advertising
Advertising
Advertising
Website
68
Three Observations for Successful Email
• It’s about APIs, analytics and getting emails to the inbox now, not about the pretty SaaS campaign management UI.
• Real-time detailed data about customer email behavior and responses is needed.
• Plan for peak volumes with email delivery service performance commitments.
69
APIs and Analytics
NOW:
Predictive Analytics Drive Targeting & Offers Today
THEN:
Human Targeting via SaaS Campaign Manager
70
What is an API?
• API is an acronym for Application Programming Interface
• An API is a set of programming instructions and standards for accessing a Web-based software application.
• API's are software-to-software interfaces, not a user interface As an end user of a system that uses APIs, applications will talk to each
other through these API's without your knowledge or a need for you to intervene
• A software company releases its API to the public so that other software developers can design solutions that are powered by its service
71
Important Considerations
• Choice of Analytics Product
Recommendations Streaming Data Predictive Analytics Custom
• Round-trip Data
• Speed of Process
• Delivery Capabilities of the Sending Engine to ISPs
• Transmission API • List API • Template API • Metrics API
Email Generation
and Sending
XLS
4 API’s You’ll Need
72
Real-Time Detailed Data: You Cant Improve What You Can’t See
• Email Delivery Data is as important as Web Clickstream Data
• Data is the Key to Analytics and Business Improvement Revenue Optimization Brand Protection – ISP Blocking & Problem Resolution Multi-Channel Engagement
73
Why Real-Time?
• In market buyers don’t wait
• Retarget them… NOW!
• Typical ESP data is delayed and summarized
74
Why Detailed Data? — Email Data Done Wrong
Summarized data can obscure insights.
75
Why Detailed Data? Email Data Done Right
Hourly details enable time delivery optimization.
76
Message Systems Data and Analytics
• Integrated data warehouse and BI tool for ad hoc analysis.
• Sensible and useful standard reports with drill down.
• Accessible details for moving data to other systems.
Data Access Method For When You
Integrated Data Warehouse and BI Want to analyze email performance in near real time
Web Hooks Stream real-time detailed data for big data analysis
Metrics API Programmatically call and extract the email detailed data you need
Log File Batch delivery of full detail
77
Plan for Peak Volumes — Managed Cloud for Data Driven Marketers
• Hosted at a Service Provider Site • Supports thousands of customers • Often utilizes shared infrastructure • Suited for information that
is not sensitive • Slow peak time performance • Pay by usage • Standard service
• Hosted at a Service Provider (Amazon) • Supports one customer • Does not utilize shared infrastructure • Suited for information that requires a high level
of security • Defined peak performance level • Fixed payment • Customizable
ESP Markting Cloud
Message Systems
Managed Cloud
78
Thank You!
Follow us on Twitter: • @messagesystems
Follow us on Linkedin:
• Message Systems
Visit Us • www.messagesystems.com
Contact Us