Big Data Event - 2013
breakfast event: Big Data 2013
The Royal Horseguards Hotel
Schedule
8:00am Welcome
8:30am  Introduction – Chris Eldridge MD (PSD)
 Guest Speaker – Mike Fishwick
(Telefonica Digital)
 Summary – Chris Eldridge
9:00am  Questions & Discussion
 Breakfast & Networking
9:45am Close
Making money from your data
Mike Fishwick
18th September 2013
Broadly covering….
 Broad business context for Telefonica
 Big Data - a few stats from mobile world
 Essential business proposition created for Telefonica
 People stuff
 Products - do they really work & can we make money
 Data Quality – A cautionary note
 Summary and final considerations
Going Digital…
 Telefonica Digital formed 18 months ago
 Consolidate and accelerate “non-Core” products and services to
the market
 Core Mobile telephony is commoditising
 Objective to become an information company
 Identified 3 key “information needs”
 BI unit for Tef Digital
 A global BI transformation programme across the OB’s
 Monetise our data assets
Food for thought….
“Big data is like teenage sex:
everyone talks about it,
nobody really knows how to do it,
everyone thinks everyone else is doing it,
so everyone claims they are doing it.”
anon
What is Big Data?
Volume
Variety
Velocity
From Dawn of Time to 2003 5 Exabytes were created; Now generated every two days
90% of mobile users keep their mobile <1 metre, 24/7
Monetising your Data
 Monetisation can take two major routes
 Internal monetisation - Customer and Operational effectiveness
 External monetisation
 Internal Monetisation
 Suggest it becomes the remit of the BI team that you have internally
 They have the technical skills to lead
 They have the science skills to analyse
 They have the commercial skills to interpret
 External monetisation
 Don’t mix it up with the internal activity
 Create it as a separate function to drive commercialisation
 Make it a part of your digital business plans
The Business Model....
Privacy is a vital to the approach
• Anonymised
• Aggregated
• Extrapolated
&
• Dispersed
Organisational Design
 A business unit with full P&L NOT sales
 Recruitment of industry specialists – retail first – to blend in
with the existing product technologists
 Investment in Data engineers and Data scientists – NOT
network engineers
 Building a skunkworks or lab function and NOT being afraid to
seek direct customer feedback as we formulate the product
 Sell with specialists NOT the Telco sales team
 Augment with partners NOT We can do it ourselves mentality
360°Retailer view – Retail Product
vision
Catchment
Location footfalls
In-Store Ffalls
Offline Checkouts
Online Checkouts
Geo-located
Online Visits
of product pages
Geo-graphical mapping
of product
Demand based on
Online Logs
Intra Store Ffalls
What our customers
already know
What we can tell
Our customers
JetSetMe – Product Overview
1. Consumer opts-
in (one-time) using
mobile number
Merchant
2. Consumer
makes purchase
at merchant with
card
3.
Authorisation
request
5. Approved
or Declined
4. Real time risk analysis performed with
addition of TEF customer geo-location data
Location updates
sent to bank fraud
system
6. Approved
or Declined
Product Roadmap
Aggregated data from
third parties – CRA for unbanked
subprime
TDI Historical data – Credit
Scoring
Personal data – Identifty Provider
(IDP)
Number reputation - Geolocation
ID&F
Datatypes/capability
JSM
Mexico
Germany
Ireland
Spain
Brazil
UK
JetSetMe is the first building block in an ecosystem of solutions that use mobile data
to address identity related opportunities. JSM addresses a global fraud opportunity
estimated to be in excess of € 6 billion.
Value proposition – (Card Present abroad)
• Costs of processing
flagged/blocked transactions
• Costs related to call center
(pre travel, to un-block)
• Number of transactions
• Avg. transaction value
• Transaction fee
• Amount of detected fraud
transactions
• Avg. transaction value
• Number of False / + declines
• Avg. declined value
• Transaction fee
Saved operation costs
Revenues from additional
card usage
Increased fraud detection
savings
Revenues from reduced
transaction declines
Value generated KPIs ImpactedFraud Losses
Globally ~€ 6.16 bn
136
DOMESTIC ABROAD
CNPCP
Fraud loss
UK/DE/ES ~ € 760M
1%
6%
81%
12%
Total value to trial client modelled at Multi millions pa
Data Quality
 A deep and challenging topic BUT in summary the following are just some of the
issues my team came across and had to address….
 The issues with data quality are NOT TO BE TRIVIALISED if one is going
beyond the network operation to one of using DATA for DATA PRODUCT
development.
 Issues relate to how the network behaves under load
 Issues of down time for maintenance.
 Referential integrity because the network engineers change the topology (and don’t tell us!)
 General failure of network devices like probes
 All this means that understanding the data is critically important
 What event types do we use
 How do we model these network behaviours
 How to deal with data losses – time series/trending requires this
 Locational Precision
 effective aggregation
 Land use and where not to put people
How was it all achieved?
 Dealing with the existing business model
 Taking a different view about organisation
 Examining the leadership model
 Throwing away the take to market model
 Leveraging new technologies (All cloud & Open source Based)
 Taking managed risks
Summary
 Moving from the unknown unknown’s to the “we now know what we don’t know”
 Investing time in understanding the source data – go beyond just enough
 Product architecture is key to agility
 Data architecture is key to flexibility
 Be clear about the debate between product and platform as the proposition
 Don’t let the Legacy mentality take over
 Get someone in who’s hand-produced the T shirt
Big Data 2013
Contact:
Chris Eldridge MD (PSD)
Chris.Eldridge@psdgroup.com
Tel: 0207 970 9700
PSD TechnologyFind us on
Facebook

Big Data Monetisation

  • 1.
  • 2.
    breakfast event: BigData 2013 The Royal Horseguards Hotel Schedule 8:00am Welcome 8:30am  Introduction – Chris Eldridge MD (PSD)  Guest Speaker – Mike Fishwick (Telefonica Digital)  Summary – Chris Eldridge 9:00am  Questions & Discussion  Breakfast & Networking 9:45am Close
  • 3.
    Making money fromyour data Mike Fishwick 18th September 2013
  • 4.
    Broadly covering….  Broadbusiness context for Telefonica  Big Data - a few stats from mobile world  Essential business proposition created for Telefonica  People stuff  Products - do they really work & can we make money  Data Quality – A cautionary note  Summary and final considerations
  • 5.
    Going Digital…  TelefonicaDigital formed 18 months ago  Consolidate and accelerate “non-Core” products and services to the market  Core Mobile telephony is commoditising  Objective to become an information company  Identified 3 key “information needs”  BI unit for Tef Digital  A global BI transformation programme across the OB’s  Monetise our data assets
  • 6.
    Food for thought…. “Bigdata is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it.” anon
  • 7.
    What is BigData? Volume Variety Velocity From Dawn of Time to 2003 5 Exabytes were created; Now generated every two days
  • 9.
    90% of mobileusers keep their mobile <1 metre, 24/7
  • 10.
    Monetising your Data Monetisation can take two major routes  Internal monetisation - Customer and Operational effectiveness  External monetisation  Internal Monetisation  Suggest it becomes the remit of the BI team that you have internally  They have the technical skills to lead  They have the science skills to analyse  They have the commercial skills to interpret  External monetisation  Don’t mix it up with the internal activity  Create it as a separate function to drive commercialisation  Make it a part of your digital business plans
  • 11.
  • 12.
    Privacy is avital to the approach • Anonymised • Aggregated • Extrapolated & • Dispersed
  • 13.
    Organisational Design  Abusiness unit with full P&L NOT sales  Recruitment of industry specialists – retail first – to blend in with the existing product technologists  Investment in Data engineers and Data scientists – NOT network engineers  Building a skunkworks or lab function and NOT being afraid to seek direct customer feedback as we formulate the product  Sell with specialists NOT the Telco sales team  Augment with partners NOT We can do it ourselves mentality
  • 14.
    360°Retailer view –Retail Product vision Catchment Location footfalls In-Store Ffalls Offline Checkouts Online Checkouts Geo-located Online Visits of product pages Geo-graphical mapping of product Demand based on Online Logs Intra Store Ffalls What our customers already know What we can tell Our customers
  • 17.
    JetSetMe – ProductOverview 1. Consumer opts- in (one-time) using mobile number Merchant 2. Consumer makes purchase at merchant with card 3. Authorisation request 5. Approved or Declined 4. Real time risk analysis performed with addition of TEF customer geo-location data Location updates sent to bank fraud system 6. Approved or Declined
  • 18.
    Product Roadmap Aggregated datafrom third parties – CRA for unbanked subprime TDI Historical data – Credit Scoring Personal data – Identifty Provider (IDP) Number reputation - Geolocation ID&F Datatypes/capability JSM Mexico Germany Ireland Spain Brazil UK JetSetMe is the first building block in an ecosystem of solutions that use mobile data to address identity related opportunities. JSM addresses a global fraud opportunity estimated to be in excess of € 6 billion.
  • 19.
    Value proposition –(Card Present abroad) • Costs of processing flagged/blocked transactions • Costs related to call center (pre travel, to un-block) • Number of transactions • Avg. transaction value • Transaction fee • Amount of detected fraud transactions • Avg. transaction value • Number of False / + declines • Avg. declined value • Transaction fee Saved operation costs Revenues from additional card usage Increased fraud detection savings Revenues from reduced transaction declines Value generated KPIs ImpactedFraud Losses Globally ~€ 6.16 bn 136 DOMESTIC ABROAD CNPCP Fraud loss UK/DE/ES ~ € 760M 1% 6% 81% 12% Total value to trial client modelled at Multi millions pa
  • 20.
    Data Quality  Adeep and challenging topic BUT in summary the following are just some of the issues my team came across and had to address….  The issues with data quality are NOT TO BE TRIVIALISED if one is going beyond the network operation to one of using DATA for DATA PRODUCT development.  Issues relate to how the network behaves under load  Issues of down time for maintenance.  Referential integrity because the network engineers change the topology (and don’t tell us!)  General failure of network devices like probes  All this means that understanding the data is critically important  What event types do we use  How do we model these network behaviours  How to deal with data losses – time series/trending requires this  Locational Precision  effective aggregation  Land use and where not to put people
  • 21.
    How was itall achieved?  Dealing with the existing business model  Taking a different view about organisation  Examining the leadership model  Throwing away the take to market model  Leveraging new technologies (All cloud & Open source Based)  Taking managed risks
  • 22.
    Summary  Moving fromthe unknown unknown’s to the “we now know what we don’t know”  Investing time in understanding the source data – go beyond just enough  Product architecture is key to agility  Data architecture is key to flexibility  Be clear about the debate between product and platform as the proposition  Don’t let the Legacy mentality take over  Get someone in who’s hand-produced the T shirt
  • 24.
    Big Data 2013 Contact: ChrisEldridge MD (PSD) Chris.Eldridge@psdgroup.com Tel: 0207 970 9700 PSD TechnologyFind us on Facebook