data science in digital marketing - forest cassidy, leadferret
TRANSCRIPT
Data Science in Marketing Forest Cassidy, CEO LeadFerret, Inc.
What is Data Science?
Wikipedia: “Data science employs techniques and theories drawn from many fields within the broad areas of:• mathematics, • statistics, • operations research,• information science,• and computer science….
…Wait, there’s more• including signal processing, • probability models, • machine learning, • statistical learning, • data mining, databases,
• data engineering, • pattern recognition and learning, • visualization, • predictive analytics, • uncertainty modeling, • data warehousing, • data compression,
• computer programming, • artificial intelligence, • and high performance computing.”
Drew Conway’s DefinitionAuthor of Machine Learning for Hackers and co-founder of DataKind
• Business intelligence is querying the past, while data science is using data from the past and present to build models to predict and/or affect the future within a scientific framework.
• Ask what if questions, then answer them based on observed and measured patterns.
What do Data Scientists Do?
• Data scientists use a scientific approach to refining big data into products, intelligence, and action.
• They use a scientific method not unlike the one you learned when creating your 7th grade science fair project.
Data Scientist Skillsfrequency per 1,000 social profiles with title =“data scientist”
Data M
ining
Data Analy
sis
Machine L
earning
Python
Statistics SQ
L
Analytics Jav
a
Algorith
ms
Big Data
Hadoop
Matlab
Predicti
ve Analy
tics C++
Database
s
Predicti
ve M
odeling
0
50
100
150
200
250
300
350
What is Big Data?
• Big data is not just about the size of the data, it’s about:– Volume– Variety– velocity
• Big data in its purest state is like noise, everywhere at once, never at rest, and continuously moving extremely fast.
Examples of Data Science at Work• Google 2009: Predicts which
regions of Mexico are at highest Ebola risk based on flu related searches
• Google’s page ranking is data science
• Google uses data science for ad targeting
• “You want to connect with this person” on social networks
• “You might also like this product” on Amazon, Netflix, Etc.
Examples of Data Science at Work
• Video games• Voice recognition• Fraud detection• Airline route planning• Price comparison sites• Delivery logistics• Coming soon: self driving
cars and robots
Examples of Data Science at Work
• Predicting which content will resonate
• Predict which social networks your prospects are likely to be listening on
• Predict which offers will work with which customers
• PPC Optimization• Retargeting Optimization
Data Science in Marketing• Churn prediction• Email campaign optimization• Custom messaging• Message timing optimization• Identifying patterns in
customers and prospects
• Start with a question and hypothesis
• Compile the data to test your hypothesis
• Maintain your control• Query the data• Analyze your results• Draw your conclusion• Take action accordingly• Tweak and start over
Identifying Patterns in Customers and Prospects
Start with question and hypothesis:• Question: Are marketers using marketing
automation software more likely to use LeadFerret.com?
• Hypothesis: Yes
Identifying Patterns in Customers and Prospectsdisclaimer: pseudo-data science
Compile the data to test your hypothesis:• Compiled 200 million+ profiles from Linkedin.• Identified marketing professionals by title.• Determined those using marketing
automation based on page scans and/or social profile content.
• Match these records to the LeadFerret user base.
Identifying Patterns in Customers and Prospects
Maintain your control(s):
• Determined the likelihood of any of the 200 million individuals being a LeadFerret user.
• (450 per million) • Determined the likelihood of any marketer in the 200
million being a LeadFerret user. • (2.7 million marketers, 12k per million use
LeadFerret, or about 32k)
Identifying Patterns in Customers and Prospects
Query the data and Analyze your results:• Determine the number of marketers using marketing
automation.• (1.05 million of 2.7 million)• Determined the likelihood of any 1.05 million marketers
that use marketing automation using LeadFerret. • (22k per million use LeadFerret, almost double the rate
of marketers in general)
Identifying Patterns in Customers and Prospects
Draw your conclusion:• Marketers using marketing automation
are more likely to use LeadFerret.
Identifying Patterns in Customers and Prospects
Take action accordingly:• We now identify marketers using marketing
automation and target them more rigorously.
Identifying Patterns in Customers and Prospects
Tweak and start over:• Now looking to see if specific marketing
automation users (Act-on, Hubspot, Marketo) have higher conversation/adoption rates than others.
Good Data science should never end.
Identifying Patterns in Customers and Prospects
Questions? LeadFerret.com866-535-3960@LeadFerret1 @[email protected]