big data analytics shaping the business of the future · big data analytics data format data type...

24
Big Data Analytics Shaping the business of the future

Upload: others

Post on 30-May-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Big Data Analytics

Shaping the business of the future

Page 2: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

• Big Data Analytics

• Accenture Point of View

Agenda

Copyright © 2013 Accenture. All rights reserved. 2

Page 3: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

• Big Data Analytics

• Accenture Point of View

Agenda

Copyright © 2013 Accenture. All rights reserved. 3

Page 4: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

What is Big Data?

Copyright © 2013 Accenture. All rights reserved. 4

• Unemployment • Interest Rate • Consumer Confidence Index • Inflation Rate • Income • Consumer Price Index

Inte

rnal

Traditional Enterprise Data

Exte

rnal

Machine Generated / Sensor Data

Social Media Data Macroeconomic / Public Data

• Customer information from CRM systems • Transactional ERP data • Web store transactions • General ledger data

• Call Detail Records • Weblogs • Smart meters • Manufacturing sensors • Equipment logs • Trading Systems data

• Customer feedback streams • Micro-blogging sites • Social Media platforms

Big data is defined by four key characteristics:

A smoothly integrated collection of diverse data-sources:

Volume Velocity Variety Value

Page 5: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Industry analysis shows that there is huge value to

be gained from Big Data Analytics

US Health Care

$300 billion value per year

~0.7% annual productivity

growth

Europe Public Sector

€250 billion value per year

~0.5% annual productivity

growth

Personal Location Data

€100 billion revenue for SPs

€700 billion for end users

US Retail

60+% increase in net

margin available

Manufacturing

50% decrease in prod dev and

assembly

7% reduction in working cap.

Copyright © 2013 Accenture. All rights reserved.

Page 6: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Dat

a V

elo

city

, Var

iety

, Vo

lum

e &

Co

mp

lexi

ty

Time

Reactive Business Intelligence

Predictive Business Intelligence

Current Generation of Big Data

Next Generation of Big Data

1

2

3

4 • Traditional structured

data (e.g. CDRs, CRM, transactions)

• Data age: Weeks to months

• Descriptive analytics (e.g. segmentations)

• Static reports

• Augmented structured data (e.g. CDRs, CRM, transactions, lifestyle, lifestage)

• Data age: Days to weeks

• Predictive analytics (e.g. forecasting, churn propensity)

• Dynamic reports

• Structured & unstructured data (e.g. location, social interaction)

• Data age: Days to weeks

• Big data analytics (e.g. social network analysis, social media analytics)

• Structured & unstructured data (e.g. clickstream data, multi-device usage data, mash-ups from multiple industries)

• Real-time big data analytics (e.g. real-time voice-to-text mining) • Next Generation underpinned by Technology innovations

Technology advances underpin how enormous data

volumes can now be processed in real-time

Copyright © 2013 Accenture. All rights reserved.

Page 7: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

What is Big Data Analytics?

Copyright © 2013 Accenture. All rights reserved. 7

Big Data Analytics is:

• a shift in the mindset of how we think about analytics as an internal component to the organization.

• a way to foster a culture around our organization that focuses on integrating data from a diverse variety of sources in such a way that drives meaningful insights in a rapid fashion.

• Hence, it leads to enhanced productivity, stronger competitive position and greater innovation.

… The way that these benefits are ensured is via simultaneously addressing:

- Enhanced productivity

- Stronger competitive position

- Greater innovation

Mana-gement Comple

xity

Missed Opportunities

Latency requirements

Missed Opportunities:

Lack of sufficient computing power

often prevent traditional analytics

tools from analyzing all data that is

available.

Latency requirements:

Lack of an effective computing

model may hinder traditional

analytics tools from taking into

account data this is dynamically

updated from a multitude of different

data-sources.

Management complexity:

Lack of a successful integration

between different organizational

departments’ data-sources typically

impedes traditional analytics tools to

infer meaningful inter-departmental

insights.

Page 8: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Gradual Progression to Big Data Analytics

Copyright © 2013 Accenture. All rights reserved. 8

• Transactions

• Call detail records

• Location

• Click-streams

• Social data

• Channel interaction

• Device usage

• Structured

• Unstructured

• Real-time

• Immense

• Real-time predictive

• Transactions

• Call detail records

• Structured

• Days to weeks

• Low to medium

• Predictive

• Transactions

• Call detail records

• Structured

• Weeks to months

• Low

• Descriptive

Last Generation Analytics

Current Generation Analytics

Next- Generation Analytics/ Big Data Analytics

Data format

Data type

Data volume

Data age

Analytics

Data +

Technology

People +

Culture

Process +

Organization Next–Gen Customer Profile

Agile Segmentation with Predictive Analytics

Social Media Analytics

Channel Usage Analytics

Social Network Analytics

Location Based Analytics

Real Time Decision Analytics

Next-Gen Analytics Capability

Accenture

Proposition

Next-Gen

Customer

Service

Next-Gen

Business

Innovation

Next-Gen

Business

Service

Page 9: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Typical Big Data Analytic Architecture

Copyright © 2013 Accenture. All rights reserved. 9

Aggregations

Pattern Mining

Transformations

Big Data Analytics

New

Features Feature

Repository

Ads

Offers

Recommendations

Propensity

Big Data

Structured Silo #1

Structured Silo #2

Structured Silo #3

Unstructured Data

Page 10: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

• Big Data Analytics

• Accenture Point of View

Agenda

Copyright © 2013 Accenture. All rights reserved. 10

Page 11: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Analysis shows that the industries that would most benefit from Big Data

analytics are Retail, Media, Financial Services and Healthcare

Value Potential of Big Data and Ease of Value Capture, Europe, 2011

Scale of

Data

Usage

High

Low

Value Potential Generated by Big Data Low High

Accommodation

& Food

Manufacturing Utilities

Health Care

Mining

Comms,

Media &

Information

Finance & Insurance

Transport & Storage

Real Estate

Government /

Public Services

Retail (Trade / Distribution)

Construction

Scientific & Tech

activity

Size of bubble represents potential economic value added by industry sector

Source: Accenture analysis based on IDC, Eurostat, Gartner and McKinsey estimates

Industry Prioritisation

1. Retail & FMCG

2. Advertising/Comms/ Media

3. Financial Service

4. Government & Transport

5. Healthcare

Copyright © 2013 Accenture. All rights reserved.

Page 12: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Building products for these industries based on geo-location use cases,

gives significant potential market revenues

• Real-time footfall analysis products can

be created from mobile network events

(e.g. connecting to a cell tower, voice,

texts), Wi-Fi data, GPS coordinates and

other forms of geo-location big data.

• Such products allow accurate and

frequent tracking of population

movements, especially with regards to

footfalls in retail catchment areas

• Retailers can use this information to

devise targeted marketing campaigns

and optimise store locations

Retail – Footfall and Segmentation

• Real-time footfall analysis products can

be created from mobile network events

(e.g. connecting to a cell tower, voice,

texts), Wi-Fi data, GPS coordinates and

other forms of geo-location big data

• This can be used for the advertising

industry to more accurately calculate

marketing ROI for the out-of-home media

channel (e.g. billboards, transport

signage)

Media – Outdoor Media Planning

• Banks are also partnering with mobile

network operators to improve their real-

time fraud detection and credit scoring

• Fraud detection algorithms can be

augmented by the location of a

customer’s mobile phone when making

online transactions

Financial Services - Fraud

• Smart Cities topic utilises multiple forms

of Big Data (e.g. M2M, street light

sensors, telematics sensors) to aid

governments in building a real-time

dynamic view of an urban population

• Big Data from telematics devices can

also enable real-time optimisation of

traffic flows in a city

Smart Cities / Traffic Optimisation

Copyright © 2013 Accenture. All rights reserved.

Page 13: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Mobile Usage Tracking

Offer Definition

Provide aggregated mobile Internet,

App, Device and Ad usage metrics

based on actual consumer usage on

the Telco network

Target Customers

Outdoor Media Platform

Offer Definition

Provide direct outdoor audience

measurement and expand into digital

signage infrastructure

Target Customers

The three “beachhead” products for Marketing and Media Sector

CPS / Sales Conversion

Offer Definition

Generate customers for advertisers

by providing “customer value” insight

and running multi-channel

campaigns

Target Customers

Outdoor Publishers

Event Venue / Properties

Outdoor & Event Marketers

Outdoor & Event Marketing Agencies

Mobile Device OEMs

Digital Publishers

Digital Advertising Agencies

Digital Marketers Digital Marketers

Copyright © 2013 Accenture. All rights reserved.

Page 14: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Big Data Applications and Solutions

14

1.Predictive Modeling

2.Data Visualization

3.Cluster Partitioning

4.Outlier Analysis

5.AB Testing

6.Markov Chains

1.Modeling true risk

2.Customer churn analysis

3.Recommendation engine

4.Ad targeting

5.PoS transaction analysis

6.Failure Prediction

7.Threat analysis

8.Trade promotion effectiveness

Indicative Applications Solution Patterns

Copyright © 2013 Accenture. All rights reserved.

Page 15: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Big Data Solution Overview

15

SENTIMENT ANALYSIS, TEXT MINING and PREDICTIVE ANALYTICS

PERSONALISED OFFERINGS

Transactional

Machine-generated / Sensor-data

Social Media

TERADATA

Macroeconomic

Social Media & External Data Internal Customer Data

BIG DATA

DATA INTEGRATION

ETL ETL

Decision Tree Customer

Micro-Segments

1. Acquire

2. Organize

3. Analyze

4. Evaluate

Web Mobile Internal dashboards and analytics

Cosine Similarity Method

Neural Networks

Predictive Analytics

Copyright © 2013 Accenture. All rights reserved.

Page 16: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Big Data requires separate and scalable technology infrastructure (e.g. Hadoop, EMR) to traditional BI to cope with scale, velocity and variations

Da

ta In

ge

stio

n

Data

Inte

gra

tion

(RE

ST

)

HBase

Hive

HDFS

CPU

Disk

Node

Rack

CPU

Disk

Node

Rack

CPU

Disk

Node

Rack

Compute/Storage

BI Analysis Advanced Analytics

Hadoop Ecosystem

NoSQL

Customer-Facing

Apps

Real-time, In-Memory

Analytics

MPP Agile

Data Marts

Map/Reduce

Illustrative Conceptual

Solution Architecture

Copyright © 2013 Accenture. All rights reserved.

Page 17: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Selection of appropriate predictive analytics techniques

Copyright © 2013 Accenture. All rights reserved. 17

Cosine Similarity Method

• Model customer behaviour using linear algebra

• For example: customers can be modelled as vectors, merchants as vector components, spending amounts as component magnitude

Artificial Neural Networks

• Neural networks are powerful machine learning algorithms that use complex, nonlinear mapping functions for estimation and classification.

• Models with more complex topologies may also include intermediate, hidden layers and neurons

Decision Trees

• Decision Trees are perhaps the most popular classification technique

• Through successive partitions of the initial population, their goal is to produce ‘‘pure’’ sub-segments, with homogeneous behavior in terms of the output

Customer Segmentation

• Multi-dimensional grouping of customers based on needs, behavior and value dimensions, according to pre-defined business objectives

• Segment profiling as input to customer strategy around value propositions, products, services , channels & experience

Page 18: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

18

Accenture Point of View

Copyright © 2012 Accenture. All rights reserved.

360-degree view of customers

Benefits Client Examples Details

1. Improved agility to respond

to competitive threats

• European mobile operator

• South-East Asian operator

Drawing and combining data from real time

feeds and traditional historical data enables

generation of real time insights, inference of

“hot spots” (i.e. current interests, new location

patterns, reactions to competitor promotions),

presentation of the right product at the right

time to each customer.

2. Richer insights from social

media

• US-based network operator

Extracting and combining social characteristics

with existing mobile behavioral knowledge, Big

Data Analytics create deeper insights to engage

and retain existing customers.

3. Differentiated user

experience • US wireless operator

Integration of data coming from IPTV set-top

boxes and voice and data usage combined with

application of basket analysis generate:

• a channel affinity map that links channels

most likely to be viewed together,

• recommendation of appealing bundles

based on viewing patterns.

Page 19: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

19

Accenture Point of View

Copyright © 2013 Accenture. All rights reserved.

360-degree view of customers

Benefits Client Examples Details

4. Identification of leaders and

followers • US – based wireless operator

Combination of social media information can

help to follow actions of super-influencers

during all stages of any product’s lifetime. This

functionality:

• adds value during early product adoption

and service take-up stages,

• promptly identifies all signs of contagious

churn,

• provides recommendations for customized

targeting (i.e. pricing, bundle of offers of

interest) to super-influencers.

5. Smarter business decisions • Spanish mobile operator Integration of call detail records, location data

and usage patterns enables location-based real

time marketing offers.

6. Tailor-made real time

recommendations for each

customer interaction

• UK mobile operator

Application of predictive analysis and historical

customer engagement rules on real time

customer interaction information enables a

client to offer appropriate real time

recommendations to each interacting customer.

Page 20: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Business Challenge

Client sought partner to:

• analyze customer patterns of behavior, • provide recommendations to customers for best actions that have

proved of help to similar customers, • provide recommendations to customers based on their upcoming key

life events, Project scoped to span from data collection and integration to modeling and

insight generation.

Integrated a wide variety of internal and external data sources in an

efficient way.

Developed sophisticated statistical modeling techniques to identify

groups of customers with similar behaviors.

Implemented cutting edge algorithms to infer insights for customers’

best courses of action.

Developed innovative approaches to provide suggestions to customers,

depending on their corresponding key life events.

Increased Sales to existing customers: Proactively contact customers based

on behavioural triggers and key life stages.

Increased retention of existing customers: Proactively target customers with

high risk of churn with specific high value services.

Increased acquisition of new customers: Provide Personalised pricing and

use social data indicators during interactions.

Accenture Contribution

Key Benefits

Illustrative Outputs

Correlation and Prediction

Analysis and Decision Making

Recommendations and Insights

20

A case study from a major financial institution

Decision Tree

Customer Micro-

Segments

Cosine Similarity Method

Neural Networks

Copyright © 2013 Accenture. All rights reserved.

Page 21: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

21 Copyright © 2013 Accenture. All rights reserved.

Improving the performance on a US online media content

provider

Business Challenge • The client is one of the leading online video streaming companies

in the US

• The available titles are currently over 8000 and are offered either

in SD or in HD format.

• The client offers titles from movie companies such as EPIX,

Lionsgate, NBCUniversal, Paramount Pictures, Relativity and

Sony Pictures

• The client was to become the leader of the market and obtain a

market share greater than that of Netflix

• The recommendations are currently static and not generated

though an automated process

• The client wanted to evaluate the increase in the market share

and bottom line an automated intelligent recommender system

can bring in.

How Accenture Helped • Accenture engaged in a PoC with the client to assess the sales

lift that ARE can achieve.

• Accenture used multiple data sources spanning different

dimensions of the customer DNA and applied user-based

collaborative filtering techniques to identify the best possible

user-movie matches

• The solution is hosted on the cloud and it operates in real time

offering targeted recommendations to the subscribers of the

customer

Outcomes • The recommendations of ARE were compared to those of the

client using Accenture’s offline validation approach and it was

found that they improve performance by 10% in terms of

recall and precision.

• Currently a live validation is performed via the email channel:

ARE’s recommendations are offered to clients and their

responses are recorded in order to assess the effect on the

performance. Most recent results present an increase of 10-

15% in views (rentals, purchases and subscriptions)

Page 22: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

22 Copyright © 2013 Accenture. All rights reserved.

Helping the Leading Chinese online retailer improve their

recommender system

Business Challenge • leading B2C ecommerce portal (35% market share) in China.

The Client has expanded beyond their core IT and consumer

electronics products, into general merchandize and books, and

has also penetrated the C2B2C channel via their public open

platform. Its leading position has attracted investors, and they are

now planning a US IPO. Its aim is to be the Amazon.com of

China. Its annual revenue was approx. 6BN USD in 2011.

• The client is now exploring alternative growth areas, beyond

category expansion. As a key strategic thrust in 2012, they seek

to grow revenues via online product recommendations and web

page optimization. The current recommender system contribution

to sales is below industry benchmarks.

• Client is growing rapidly: Number of transactions double every 5

months and

• Long tail: 80% of their products account for only 5% of their sales

How Accenture Helped • Accenture is helping the client improve its recommendation

capability, with a focus on driving transactions via better use of

data and improved algorithms, leveraging Big Data analytics and

web page optimization.

• Accenture used a global team consisting of experts in machine

learning and big data from Adelaide, Athens, Beijing and San

Francisco to deliver results for Implementing our

recommendation engine algorithms (ARE)

Outcomes • Accenture developed approximately 40 machine learning

algorithm variants to give us the results.

• In online testing ARE outperformed the internal

recommendation engine of the client by up to an estimated

30%

• This is translated to revenue uptick for the client, which is

estimated to be initially up to ~$100M USD per year (we

expect this to increase over time).

• We have also applied out robust repeatable agile testing

platform to rapidly prototype algorithm performance. The test

platform logic is derived from the field of information retrieval

(IR) where we calibrated algorithm performance according to

their impact on precision and recall.

Page 23: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

23 Copyright © 2012 Accenture. All rights reserved.

Leading US Discount Coupons Company

Business Challenge • The client is a leading coupons company in the US offering

discount coupons to consumers. The discount coupons are

issued by manufacturers who want to use this channel to

increase their sales.

• The customer is one of the leading players in the coupons

industry and they want to increase their market share by applying

intelligent, real time automated targeted methodologies

• The client is affiliated with many grocery stores chains all over

the country

• The challenge is to combine heterogeneous data from many

different sources and to generate insights and provide

personalised coupon recommendations to customers using

various means: web site, digital receipt, emails, etc.

How Accenture Helped • Developed an in-house big data platform that collects

transactional data from affiliated grocery stores , processes them

and generates automated personalised recommendations using

ARE

• Accenture used a global team consisting of experts in machine

learning and big data from Adelaide, Athens, Chicago and San

Francisco to deliver results for Implementing our

recommendation engine algorithms (ARE)

• ARE is estimated to process 2-3 TB on a daily basis in order to

deliver updated, real-time recommendations

Outcomes • ARE has been empowered with new algorithms that cater for

the peculiarities of the coupons industry. ARE’s algorithms

recommend not only coupons to customers, but also products

for which new coupons should be issued

• The site has been re-organised and coupons are presented

to customers based on the corresponding customer to

coupon estimated interest

• During the first months of operation the coupon activation

rate has increased up to 20%, depending on the product

category

• Identification of most desired/recommended products

contributes in the design of more targeted coupons.

Page 24: Big Data Analytics Shaping the business of the future · Big Data Analytics Data format Data type Data volume Data age Analytics Data + Technology ... Accenture analysis based on

Copyright © 2013 Accenture. All rights reserved. 24

Thank you!