Brighton seo april 2015 lynchpin. segementation clustering presentation. creating clever clusters

Creating Clever ClustersSegmentation for Profit

Andrew HoodManaging Director

@lynchpin

www.Lynchpin.com

What is Marketing?

13 April 2015 © Lynchpin Analytics Limited, All Rights Reserved 2

Satisfying needs and wants

through an exchange process

Philip Kotler

“”


The management process responsible for

identifying, anticipating and satisfying

customer requirements profitably

Chartered Institute of Marketing

What is Marketing?

“”


Marketing is now as much science as art due to the confluence of two

simultaneous revolutions.

The first is technological advances in platforms like Adobe Marketing

Cloud, that allow marketers to gather, analyze and elegantly visualize

large and complex data sets to gain customer insights.

The second is the explosion of digital devices and channels - from smart

phones to social media - which are the common currency of digital

natives whose lives are transacted in pixels.

Michael Schinelli, Chief Marketing Officer at the University of North Carolina’s

Kenan-Flagler Business School

“

”

2014?...


It’s more important to

REACH the people who COUNT

than to

COUNT the people you REACH

The Segmentation Principle

Segmentation Sophistication


Just eat it: we know what’s good for you.

Fun for all the family. (If you all fit these 4 buckets)

Completely personalised.(If you play guitar for Blur)

GA Value Segmentation

Average Transaction Value only defines the immediate value of a single session

Value Segmentation looks at the longer term value

of a user (customer)



• Choose some buckets (max 4)


Segment No Value Low Value Medium Value High Value

Definition No Purchases Exactly 1 Purchase

Exactly 2 Purchases

3 or More Purchases


• Apply segments to reporting


Digital Segmentation Party

One of richest sets of

behavioural data

Some of the least sophisticated segmentation

on the planet


Key Challenge:

Think of an interesting questionThen try again…And again…

Direct Mail Segmentation Party

Limited behavioural dataNice clustering

algorithms


Key ChallengeGetting the data to model

Clustering (Segmentation Modelling)


Cluster analysis or clustering is the task of grouping a set of objects in such a way that objects in the same group (called a cluster) are more similar to each other than to those in other groups (clusters).

Real Advanced Segmentation

• Unsupervised learning– The algorithm helps find

relationships/outcomes you didn’t already know about

• Anchored on profit– Not a value hunch

• Statistically based– Extracts the signal from the

noise


Creating Clever Clusters


BEHAVIOURAL

DEMOGRAPHIC

VALUE

PSYCHOGRAPHIC

GEOGRAPHIC

OFFER BASED

BENEFIT BASED


Let’s talk money

Example: Car Rental


High ValueHigh Potential

High ValueLow Potential

Low ValueLow Potential

Low ValueHigh Potential

Value

Potential

Customer Rental

Records

Current ValueRental revenue vs cost over past 12 months

Expected ValueBasic model based on recency / frequency

Example: Car Rental


Clustered By

Membership schemeAgeGenderCar group rentedPickup locationsRental durationTime of yearBusiness / leisure profile…

“High value customers tend to rent car groups K, F, L, Z, older demographic, less corporate, different pick-up/drop-off locations”

“Frequent renters tended to be short haul for business and standard domestic, male, aged 36-56 ”

High ValueHigh Potential

High ValueLow Potential

Low ValueLow Potential

Low ValueHigh Potential

Value

Potential

Example: Car Rental

Target relevant offers to encourage low value / high potential group


Focus upgrade offers and also loyalty / retention offers to high value / high potential group

Marketing options

Outcome£6m incremental revenue over 18 months

Example: Home Insurance

Have a Policy

Cancel

Renew


Have a Policy

Cancel

Renew


• Length of relationship (summarised transactional data)

• Satisfaction level (scores from research)• Number of quotes• Age of policy• Level of cover• Additional covers taken• Price• Payment method• Number of claims• Demographics• Number of products held• Ratings district• Home details (e.g. year built)• Financial position (e.g. overlay of Experian

data)



• Length of relationship (summarised transactional data)

• Satisfaction level (scores from research)• Number of quotes

• Age of policy• Level of cover• Additional covers taken• Price• Payment method• Number of claims• Demographics

• Number of products held• Ratings district• Home details (e.g. year built)

• Financial position (e.g. overlay of Experian data)

Have a Policy

Cancel

Renew


Test and target retention campaigns on segments with the highest likelihood of success

E.g. targeted renewal discounts

Identify high value segments with high probability of lapsing and focus efforts on them

E.g. getting onto direct debit

Marketing Actions

Outcome30% increase in retention.

Analytics Ecosystem


Analytics Ecosystem


Analytics Ecosystem


Analytics Ecosystem


Digital Clustering Opportunities



Useful Tools. Not this!

Tool Time


Rstudio http://www.rstudio.com/

Rattle http://rattle.togaware.com/

http://www.rstudio.com/

http://rattle.togaware.com/

Tool Time


KNIME http://www.knime.org

http://www.knime.org/

Handy Tool links

• R Integrations

– http://skardhamar.github.io/rga/

– https://code.google.com/p/r-google-analytics/

– http://randyzwitch.com/rsitecatalyst/


https://code.google.com/p/r-google-analytics/

https://code.google.com/p/r-google-analytics/

http://randyzwitch.com/rsitecatalyst/

Stats Tutorials

– http://www.autonlab.org/tutorials/

– http://rattle.togaware.com/rattle-examples.html

Handy Tutorial links

http://www.autonlab.org/tutorials/

http://rattle.togaware.com/rattle-examples.html

Andrew HoodManaging Director

@lynchpin

www.Lynchpin.com

Cheers Folks

www.linkedin.com/in/andrewhood

Brighton seo april 2015 lynchpin. segementation clustering presentation. creating clever clusters

Data & Analytics

lynchpin analytics limited

rights reserved207

rights reserved307

rights reserved407

rights reserved5

rights reserved2satisfying

rights reserved4marketing

rights reserved1what