With rich sources of real time information available at their fingertips, customers expect businesses to understand their unique needs in the context of their current situation. However, IT leaders face several challenges that inhibit their ability to take advantage of the entire pool of data that is available to them. These challenges include highly fragmented data in incompatible formats, managing exploding social media and telemetry data, and the increasing amount of customer data residing outside the corporate firewall. Learn about the tools and best practices that your peers in market leading, global organizations are utilizing to sharpen their contextual awareness of their customers. Find out how they are competing more effectively, optimizing customer interactions across channels, managing risk exposure and increasing the efficiency of their business operations with a well-defined customer information management strategy.
11. Multi-dimensional views:
Financial Services ā Case In Point
Payment Graph (e.g. Fraud Detection,
Spend Graph (e.g. Org Drill thru,
Credit Risk, Analysis, Chargebacksā¦)
Product Recoās, Mobile Payments)
Asset Graph (e.g. Portfolio
Master Data Graph (e.g. Enterprise
Analytics, Risk Management,
Market & Sentiment)
Collaboration, Corporate Hierarchy,
Data Governance)
11
12. Madoff Fraud: WSJ
The Case for Data Governance
by Penny Crosman
JAN 8, 2014
Data locked in silos and the lack of a common customer identifier that could link accounts were
to blame for JPMorgan Chase's failure to identify Bernard Madoff's massive fraud, according to
an article in Wednesday's Wall Street Journal.
(Madoff, who was arrested in 2008, stole about $18 billion from clients, sending them fake
monthly statements reflecting fake trades, assuring customers they were getting high returns
when in fact their money was gone.)
Poor Data Management Blinded Chase to
Madoff Fraud: WSJ
Madoff Investment Securities maintained several linked checking and brokerage accounts at
JPMorgan Chase, its primary bank, for 22 years. The bank structured and sold investment
vehicles tied to the firm's purported returns. The bank has agreed to pay $2.7 billion in fines to
the federal government for failing to report warning signs of Madoff's scheme.
"Despite recognizing
by Penny Crosman suspicious activity in its U.K. unit in 2008 ā and notifying U.K. regulators
that Mr. Madoff's returns were 'too good to be true' ā the bank didn't notify its own U.S.-based
AML staff or American authorities. AML experts say that JPMorgan's anti -fraud systems should
JANhave2014
8, automatically flagged Madoff accounts across the company ," the paper reports. In one of
the terms of the bank's settlement, JPMorgan has agreed to continue reforms of its Bank Secrecy
Act/Anti-Money Laundering compliance program.
Customer data that's strewn across a company and not linked has been a problem that has
plagued large banks for many years. A London division of a bank could have no idea of the
Data locked customer in Newthe lackexample, creating fraud as well asidentifier that could link accounts were
activity of a in silos and York, for of a common customer customer service
to blameShortly before the financial crisis, several to identify Bernard Madoff's massive fraud, according to
issues. for JPMorgan Chase's failure large banks appointed C -level data
management chiefs (called chief data officers)
an article in in which all accounts, transactions and had them start related to unified customer data
Wednesday's Wall Streetand other activity creat ing a customer could be
Journal.
warehouses
gathered in one place. Bank of the West recently completed such a project.
(Madoff, who was arrested in 2008, stole about $18 billion from clients, sending them fake
During
financial crisis, these
multi
projects with an
put aside.
monthly thewith the dust reflecting fake trades, assuring customerswereto customer getting high returns
statements settling, alarge,banks -year been turning theielusive ROI they were
Recently,
few
have
r attention again
data management.
when in fact their money was gone.)
But software can only do so much. The other side to this is that in Manhattan U.S. Attorney Preet
Madoff Investment Securities maintained aseveral linkedignorance is described.
Bharara's criminal charges against JPMorgan Chase, pattern of willful checking and brokerage accounts at
Time and Chase, according to the U.S. for 22 office, The had strong reason to
JPMorgan time again,its primary bank,Attorney's years.the bankbank structured and sold investment
vehicles tied to the firm's purported returns. The bank has agreed to pay $2.7 billion in fines to
the federal government for failing to report warning signs of Madoff's scheme.
"Despite recognizing suspicious activity in its U.K. unit in 2008 ā and notifying U.K. regulators
that Mr. Madoff's returns were 'too good to be true' ā the bank didn't notify its own U.S.-based
AML staff or American authorities. AML experts say that JPMorgan's anti -fraud systems should
have automatically flagged Madoff accounts across the company ," the paper reports. In one of
the terms of the bank's settlement, JPMorgan has agreed to continue reforms of its Bank Secrecy
Act/Anti-Money Laundering compliance program.
Customer data that's strewn across a company and not linked has been a problem that has
plagued large banks for many years. A London division of a bank could have no idea of the
12
13. Data Governance:
In Service of the Business Process
ā¢ Limited to non-existent support
for roles, responsibilities, and
processes between the business
and IT
ā¢
ā¢
ā¢
ā¢
ā¢
KPIs tied to process
Monitor trends over-time
Enable business stewardship
Embedded workflows/exception mgmt.
PII data anonymized
13
16. Combination of Traditional, Social Network
& Spatial Analytics: Robust Context to the
Knowledge Graph
ā¢ Who is a high spender?
ā¢ What is their propensity
to buy?
ā¢ Is the customer within my
pre-defined Geo-fence?
ā¢ How does it influence my
marketing offers?
ā¢ Who is both influential in their community
& a high spender?
ā¢ Which products would customers prefer that
others ālikeā them have purchased?
16
17. Customer Information Management Best
Practices
Use Knowledge Graphs as the intuitive & agile
way to organize complex customer data
Establish process-centric data governance
Look beyond traditional analytics to build
robust contextual profiles
17
Customer Information shifts from being locked up in silos/RDBMS to constantly evolving Knowledge Graphs of customer profiles that democratizes customer information across the enterprise, for consumption by anyone, anywhere, anytime (within established governance policies). CustomerInformation shifts from aggregated views to contextualized profiles, especially as data volumes grow exponentially, coming from new channels requiring organizations to invest in analytics to drive relevancy through contextual awarenessCustomer Information shifts from real-time request-response to extreme processing as data streams into the enterprise through new sources (social networks, internet of things) requiring organizations to re-think their data and information architectures.
capture and evolve datamodels based on real-world complex relationships that may span processes, interactions, hierarchies,roles and domains, and extract actionable insight to drive business outcomes.High performance queries on complex, connected dataProvide Multi-dimensional views vs. single viewsAgileEasier to evolve modelEasier to capture adhoc relationships
Payment Graph (e.g. Fraud Detection, Credit Risk, Analysis, Chargebacksā¦): Detection tools such as online card validation numbers (CVNs) or SecureCode, electronic fingerprint matching and customer behavior pattern analysis are all increasingly being deployed to detect fraud. However, more sophisticated means are required in order to combat the increasingly collaborative approach used by fraudsters. An indispensable method employed by investigators to gather fraud intelligence is the analysis of fraud cases successfully solved, and extrapolation of the results to prevent future fraud. This is achieved by building up a visual picture of behavior patterns which can be easily shared for fraud intelligence and prevention purposes. For example, many fraud events involve attempts to create distance between the perpetrator of the fraud, and the individuals or organizations carrying out the transactions. This indirection often involves one, two, or even three levels of separation between the parties. Where Spectrum Graph Hub excels is in allowing you to make these connections. By storing transactions and parties in the system, you can often infer relationships and walk those relationships in order to detect potential fraud patterns. In cases where you have additional information about the parties (such as organizational affiliation; personal, professional, or family relationship; address history; etc.), this information can be used to further hone oneās detection abilities. This can be done by connecting a transaction with a known fraudster; or by connecting a pattern of transactions with patterns that are known to be fraudulent. Ā Spectrum Graph Hub excels at walking relationships--it can traverse millions of relationships in under a second. It is also finely tuned for local pattern recognition. Finally, as a graph database, it supports the visualization of those patterns, making it very easy for fraud technicians, analysts, and developers to pinpoint instances of fraud.Ā Uses of Spectrum Graph Hub for fraud detection normally fall into one, two, or all three of the following categories:Real-time fraud detection systems, which flag events as fraudulent based on an already known pattern.Off-line research, where patterns are uncovered and business rules established, for use in the real-time detection engine (above).Near-real-time fraud analysis, where human beings screen transactions that have been flagged as potentially fraudulent, and use various techniques, including graph visualization, to determine whether a particular event is fraudulent.Customer Graph (e.g. Org Drillthru, Product Recommendations, Mobile Payments, Etc.): American Express is accumulating a āspend graphā database of its customersā transaction data to help brands customize their mobile marketing to consumers. The idea is to create a ācommerce IDā for consumers, which would act a sum of their transactions and brands they are loyal to. Partner brands of American Express could potentially use this data to target smartphone users with offers and ads when they are in the vicinity of a retail store, or elsewhere on the internet where they might have an intention to make a purchase.Entitlement Graph (e.g. Identity & Access Management, Authorization, etc.): We are seeing a lot of large corporations turning to graph databases to manage entitlements and complex access and authorization networks. Telenor, for example, replaced the relational database-based access control element of one of their software-as-a-service offerings with a graph database, and in so doing reduced the latency of some of their complex queries, which span millions of entities and relationships, from several minutes to several milliseconds.Asset Graph (e.g. Portfolio Analytics, Risk Management, Market & Sentiment, etc.): Highly diversified investment portfolios may reduce risk exposure, but also make it much more difficult to track profit/loss and measure staff performance. Presenting portfolios, transactions and staff activity in unified network visualizations give banks better understanding of their risk, staff performance and dependence structures within financial assets. Banks rely on resilient and reliable IT infrastructures to power their payment systems. Just a few minutes of downtime can cost millions of dollars in lost transactions and customer dissatisfaction. Network visualization can be used to achieve more secure and reliable IT systems.Master Data Graph (e.g. Enterprise Collaboration, Corporate Hierarchy, Data Governance, etc.): Banks have unparalleled insight into how customers spend money. However, data silos often prevent them from using this data for successful up-selling. Using network visualization to achieve a ā360-degreeā view of customers helps reduce customer churn, increase customer up-sell and allows banks to create offers with impact. It is ironic that relational databases are not well suited for managing relationships. But Graph Databases were built for just this sort of complexity and understanding.
capture and evolve datamodels based on real-world complex relationships that may span processes, interactions, hierarchies,roles and domains, and extract actionable insight to drive business outcomes.High performance queries on complex, connected dataProvide Multi-dimensional views vs. single viewsAgileEasier to evolve modelEasier to capture adhoc relationships
Visualization with an eye to enabling the data stewardSupport information governance processes through simplified viewsDiscovery capabilities for exploring data modelsData quality and data governance context ā how can data visualization help my business?
We provide a variety of views for visualizing and monitoring data quality.
Main point: A new generation of analytics to deepen your understanding of the customer based on a contextually relevant view by combining social network analysis & spatial analysis with traditional analysis.Who is a high spender?What is their propensity to buy?<click>Answering the where? What do the Location characteristics tell me?<click>Which person is both influential in their community and a high spender? What products would my customers like that they donāt have yet, but others with shared interests do?