Unlocking the data possibilities of Big Data presentation shared at the Big Data / Internet of Things Conference Board Conference June 25-26, 2015
http://www.pwc.com/us/en/analytics/big-data.jhtml
3. PwC's Global Data & Analytics Survey: Big
Decisions™
• Big decisions have a big impact on future profitability; however, more
big decisions are made opportunistically than deliberately
>$1bn
• Highly data‐driven companies are three times more likely to report
significant improvement in decision making, but only 1 in 3 executives
say their organization is highly data‐driven.
• The majority of executives rely more on experience and advice than
data to make business‐defining choices.
• Many executives are skeptical or frustrated by the practical
application of data and analytics for big decisions, especially in
emerging markets.
62%
3X
Data
Quality
Usefulness
1,135 senior
executives
interviewed
from across the
world
representing a
total of 18
industries
where majority
(74%) of
companies
reported annual
revenues last year
of at least $1bn
Source: PwC’s Global Data & Analytics Survey: Big Decisions™
5. Big impact on future profitability
Big decisions have a big impact on future profitability; nevertheless, more big decisions are made 'in the
moment' (either reactively or opportunistically) than deliberately.
4%
9%
15%
18%
25%
30%
Mandatory
Reactive
Experimental
Deliberate
Delayed
Opportunistic
Motivators of Big DecisionsImpact on Profitability
< $10m
$10m to
$100m
$100m to
$1bn
$1bn to
$5bn
>$5bn
NA
1 in
3 33%
Source: PwC’s Global Data & Analytics Survey: Big Decisions™
6. Big improvement on decision making
Highly data‐driven companies are three times more likely to report significant improvement in decision
making, but only 1 in 3 executives say their organization is highly data‐driven.
Significant Improvement?How Data Driven?
0% 10% 20% 30% 40%
Highly data-driven
Somewhat data-driven
Partly data-driven
Somewhat
data‐driven
Highly
data‐
driven
Partly
data‐
driven
Other
43%
14%
15%
3X
1 in
3
Source: PwC’s Global Data & Analytics Survey: Big Decisions™
7. Organizations that delay starting the Big Data
journey risk being leapfrogged by more data-
savvy competitors
58%
PwC’s Digital IQ Survey 2014 respondents who
indicated transitioning from data to insight is a
major challenge
8. #2 How can your organization adapt and
execute?
9. Many organizations face challenges in adapting
to the recent trends in the Big Data landscape
Information explosion due to Digitization, Internet of Things and external data have increased the number
of data sources, volumes and complexity available for analytics to achieve competitive advantage
Proliferation of commoditized technologies to enable speed and sophistication of high volume data
processing and analytics have contributed to a complex technology landscape
Enterprises have to balance near‐term and long‐term goals while enabling data and analytics capabilities in
an agile manner, to realize iterative business value before committing to long‐term investments
Data monetization strategies are increasingly adopted among competitors across many industries to
develop innovative products/services and generate new revenue streams
10. Big Data into Big Revenue – Journey Building Blocks
DISCOVERYDISCOVERY INSIGHTSINSIGHTS ACTIONSACTIONS OUTCOMESOUTCOMES
Discover value in your
internal and external data
Apply analytic techniques on
internal and external data
for tailored, value creating
insights
Make decisions; deliver
quick wins; build operational
capabilities to enhance
products and services
“Observations to Information” “Information to Insights” “Insights to Actions” “Actions to Outcomes”
OPERATING MODELOPERATING MODEL
Test and learn; link insights
and actions to financial and
operational metrics;
enhance shareholder value
SHAREHOLDER VALUE CREATION
How can companies adapt and execute? The
‘DIAO’ mindset
11. Discovery: Observations to Information
D1. Idea IntakeD1. Idea Intake
• Develop a process to intake and build a pipeline
of ideas on improving business decisions with
data and analytics, both from internal
organization resources and external partners
D2. Idea QualificationD2. Idea Qualification
• Qualify ideas based on potential business value
(financial, operational, risk or quality metrics)
D3. Identify Data AssetsD3. Identify Data Assets
• Identify internal/external data sets required to
unlock the value out of the idea; e.g., data sets
may cover a broad spectrum of domains,
namely customers, products, services, sensors,
demographics, social media
D4. Platform, Tools and Infra.D4. Platform, Tools and Infra.
• Develop ‘data lake’ architecture; make
technology decisions and operationalize
infrastructure to capture and store data assets
from internal and external sources
DD II AA OO
12. Insights: Information to Insights
I1. Analytics TechniquesI1. Analytics Techniques
• Categorize the type of analytics techniques
(forecasting, clustering, regression, time series,
machine learning, etc.) required for the ideas
and map analytics tools to purpose
I2. Analytics ArchitectureI2. Analytics Architecture
• Develop the ‘right fit’ architecture with tools to
enable a rapid prototyping environment.
Consider scalable in‐memory analytics and
visualization tools as core components
I3. Ideation SandboxesI3. Ideation Sandboxes
• Develop a holistic ideation sandbox strategy
and tool environment to empower practitioners
in their data discovery process. Consider cloud
models and tools available as an enabler
I4. Process AgilityI4. Process Agility
• Develop efficient processes in the discovery
lifecycle which promotes agility and eliminates
administrative bottlenecks; e.g., a self‐service
sandbox provisioning model
DD II AA OO
13. Actions: Insights to Actions
A1. Decision ModelA1. Decision Model
• Define decision models and rights that
categorize and specify the decisions that get
made, insights, options, subsequent actions
and potential for automation
A2. AutomationA2. Automation
• Integrate and automate decisions made from
models with company’s existing business
processes, operations and technology in real‐
time; e.g., Are your sales processes ready to
handle the predicted cross‐sell / up‐sell
scenarios?A3. Embed resultsA3. Embed results
• Embedding decision results into new products
and services design could be a game changer
and avenue for many organizations to add
shareholder value
DD II AA OO
14. Outcomes: Actions to Outcomes
O1. Impact LinkageO1. Impact Linkage
• Establish tighter link and integration between
insights generated, actions taken and impact to
financial, operational and risk metrics
O2. Monitor and ObserveO2. Monitor and Observe
• Monitor any deviation from the expected outcome
of predicted business impact, filter external factors
(e.g., inflation, dynamic market trends) to
measure effectiveness of management decisions
O3. Test and LearnO3. Test and Learn
• Foster ‘test and learn’ culture where people
can implement change in decisions and actions
in a limited form, observe the results, and
change the model to reflect reality
O4. Data MonetizationO4. Data Monetization
• Explore monetization strategies with the
insights gained as an additional revenue source
for the organization; e.g., licensing fee for
aggregated data sets as an event indicator
DD II AA OO
15. Four Primary Types of Operating Models
• Team typically reports to the CIO
and provides data delivery,
reporting and business
intelligence services
• Investment focused on
Infrastructure and Tools
• Primary focus on acquiring,
storing, managing and reporting
the information as opposed to
developing deep analytic
modeling skills
• Less focus on innovation and
usage of 3rd party data
Information EnablerInformation Enabler
• Team reports to functional
leaders (e.g., Marketing, Sales,
etc.) that build targeted data
marts and analytic models to
improve functional performance
• Relies on the services provided
in the “information enabler
model” as well as their own
specialists to enable data
capabilities
• Heavy focus on 3rd party data
and exploring new analytic
techniques and tools
FunctionalFunctional Cross FunctionalCross Functional
• The group reports to business
unit or P&L owners (e.g., chief
digital officer, VP of
online/mobile) and creates
value by embedding data and
analytics‐driven offerings into
new or existing products and
services
• Focus is on the impact to
revenue, profit and shareholder
value growth
• Investments are made in
innovation and 3rd party data,
as well as deep analytic models
Business Unit
/ P&L Owner
Business Unit
/ P&L Owner
• Team reports to a cross‐
functional business role (e.g.,
CFO, COO) to deliver cross‐
functional analysis to support
strategic, financial and
operational decisions that span
multiple functions
• Investments are made in
innovation, 3rd party data sets
and tools, as well as proprietary
analytic models
• Skills include data scientists and
deep quantitative experience
The Data and Analytics Operating Model Determines Your Speed to New Value
Operating with a DIAO mindset requires
rethinking the data and analytics operating model
18. Make space for profits!
Consumer product goods company
• Inventory stock out average of 13% vs.
8% industry average
• Difficulty accurately predicting demand
across a distribution network of over
1000 area sales managers
• Supply chain challenges: backroom
inventory at 24% of volume – and rising
• Sought a demand driven inventory and
shelf optimization system that provided
accurate demand forecasts for use by
sales managers on a daily basis
• Design and execution of a pilot initiative
— Time series analysis models predict
demand at a store SKU level
— Forecasting variables include effects
of price, promotions, seasonality,
product sales velocity, day of the
week , delivery constraints and others
• Develop business case, design, develop,
roll out and implement solution
• Measure performance and results
• Out of stock conditions reduced on
average to 6%
• Improved cash flows due to reduced
back room inventory
• Projected $30m EBITDA contribution a
year.
Business Issues Action Results
19. Complete
forecast creation
for 3 wks
Make space for profits!
Big Data, analytics and decisions
1. DATA
1 Classification of products
based on average volume sales
Complete
Sales Data
High Volume
Items
Low Volume
Items
2 Classification of high volume items based on formats and volume of sales
2. ANALYTICS
3
Low Volume
Item Forecast
Forecasting for
low volume
items based n
the sales of last
8 weeks
4 Input sales data
in respective
time series for
every
combination
+ Complete
Price Information
(past 2 years)
5
Forecast calculation for
every sales‐item
combination based on
best time series model
6
High Volume
Item Forecast
7
Correct the
sales time
series based
on discount
data to get
base demand
3. DECISIONS
+Daily Sales
Information for
past 12 weeks
8
Splitting the
weekly forecast
9
Handheld
Area Sales
Manager
Updated
Forecast
Make overrides
if necessary
NEXT DAY DELIVERY
20. New revenue from where streets have no names
B2B specialty pharmaceutical sector
• Flat revenues over three years
• Recent 16% reduction of sales force
• Inefficient sales force optimization,
workloads rewards and compensation
• Poor employee morale
• Big Data pilot using advance analytics
• Development of a customer value
assessment framework
• Identification of high value customer
segments
• New targeting strategy
• Redesign of sales territories
• Reprioritization of sales resources and
deployment
• Development of a business case for 2012
revenue impact
• 5%‐7% revenue lift
• More efficient sales force (16% leaner)
• Improved insight into high potential
accounts
Business Issues Action Results
21. New revenue from where streets have no names
Customer segmentation and sales targeting
1. DATA 2. ANALYTICS
3. DECISIONS
Master Data
Data integration…
Patient Data
• Office location
• Visit frequency
• Services used
Consumer Data
• Demographic
• Insurance
• Lifestyle
Sales Data
• Sales agent location
by market / territory
• Product revenue by agent
/ market / territory
Customer segmentation…
Who to target?
Value based segmentation techniques determine
• High potential customers
• Best potential customers
When to target and where?
An independent RFM process was run to segment priority customers by:
• Average spend per prescription refill
• Average time between prescription orders
• Transactions by zip code
Redesign sales territories and sales force deployment….
Define
Principles
1
Define
Constraints
2
Perform
Optimization
3
Calculate
Metrics
4
Target Markets &
Customers
5
Define workload,
potential and
performance based
principles to act as
territory balancing
criteria
Build constraints to
meet
specifications(e.g.
balanced workload)
and maintain
geographic continuity
Use statistical tools
and algorithms to
meet design
objectives and
constraints
Calculate and forecast
key metrics of new
territories
Generate customer
level targeting lists.
Develop a visual
representation of
targeted and omitted
customers on
potential map
22. Consumer insights journey
Global retailer company
• Goal was to enhance how they spend
$400m in customer based marketing
across multiple channels annually to get
the largest return on our investment
(higher sales, margins)
• Biggest foundational challenges
identified was the number of Customer
Data silos, quality of data and analytics
around the enterprise causing customer
disappoints and hurting sales (e.g.
thanked 20,000 customers for purchases
they never made, misplaced loyalty
points in other customer accounts)
• Company was spending $4‐5m annually
in marketing messages and campaign
activities with improvement
opportunities
• Funded an enterprise wide initiative for
Customer Data to
— Integrate the customer data across
multiple channels – stores, online,
mobile under one analytics repository
— better understand the transactions
and interactions of all its customers
across all of its channels by the usage
of analytics (Customer Identification,
Segmentation, Clustering)
— Use the insights generated using
analytics to better target customer
based on their preferences. Integrated
the results into 1‐1 marketing and
personalization initiatives like the
online recommendation engine
• Increased gross margin (GM) per
customer by capturing 10% more
margin for 5% of identified customer
across each of our value tiers
• Improved efficiency in the TV/Digital
marketing spend, duplicate mail savings
and identified cost take outs of ~5m in
annual budgets
• Increased offer conversion rate by 10%
on a quarterly basis
• Projecting hard benefits in the range of
50 – 55m this year in Net Operating
Profits as a cumulative effect of the
customer data program
Business Issues Action Results
23. Consumer insights journey
Big Data, analytics and decisions
1. DATA 2. ANALYTICS
3. DECISIONS
1 Single view of customer transactions and interactions for
products and services across all channels
Stores
Online
Mobile
Single View of
Customer
2 Created multiple rich segments of customers integrated across channels based on a set
of key drivers through segmentations and clustering techniques to enable personalized
targeting of offers and promotions
Customer
Engagement
Customer
Value
Customer
Behavior
Demographics
Best Customers
Important
Opportunistic
Uncommitted
Price Sensitive
Quality before
Price
Product based
promotions
New Customer
Most Loyal
Retained
/Reactivated
Prefer online
shopping
Buy online,
pickup store
Filtered a sample of most loyal members who
mattered and shopped online
Decision
/Personalization
Engine
Passed the insights to the
personalization/ decision engine
feeding the online and mobile
portals
Mobile
Web
Shopping Portal
3 Presented relevant offers, recommendations. Increased
conversion rate, profits and customer delight