Contenu connexe Similaire à Solving the Data Challenge in Compliance (20) Solving the Data Challenge in Compliance2. Hanjo Seibert
Accenture – MD,
Compliance Lead
hanjo.seibert@accenture.com
+49 15157142279
Isabelle Flückiger
Accenture – MD,
Data Analytics Lead
Isabelle.flückiger@accenture.com
+41 79-654-0726
SPEAKERS
2Copyright © 2019Accenture.All rights reserved.
3. TODAY: MISSION IMPOSSIBLE
3
The banking industry needs to identify moneylaunderers, terroristsand fraudsters,
howevertraditional approaches struggle to “discriminate”well
Without access to relevant data, everyone looks the same
Client #2 Client #4
Client #5
Client #1
Client #3
Client #6
Copyright © 2019Accenture.All rights reserved.
4. IMAGINE …
4
How much more is possible by bringing integratingdata and applying new analytical approaches
With all our data in the right place we are able to “connect the dots”
Client #2 Client #4
Client #5
Client #1
Client #3
Client #6
Copyright © 2019Accenture.All rights reserved.
5. 5
Data sharing
acrosssilos
... All the information is in the
right place, but...
Shared & integrated analytics
... It only comes together with
the right analysis being
performed...
Data & insights driven business
decisions
... With business learning how
to take insights driven
decisions...
WHAT IT TAKES
X
Copyright © 2019Accenture.All rights reserved.
6. SHARING: WHAT IT CAN LOOK LIKE
One centralplatform for E2E processes
Business Rules
Data Quality
Data Model
INTEGRATED ANALYTICS PLATFORM
Data Layer 1
Data Layer 2
Data Layer ..
SOURCES
Ext.data
sources
Internaldata
architecture
Static data
Products
Transactions
SKILLED TEAM
Governance & security
Usecases
News
Authorities
Lists
…TM KYC TAX RM Filt. …
Streamlined processes
AI & ML technology Integrated Regtech
Advanced analytics
Drivers
• Business SME
• Business expert
• Data scientist
• IT specialist
• ...
Up-skilled, integrated
team
Integrated
data set
Intelligence &
analytics led
6Copyright © 2019Accenture.All rights reserved.
7. SHARING: BENEFITING THE ENTIRE ORGANIZATION
Increased efficiency
and reduced cost –
enabling savings of >
50 %
Avoided duplicate data
sourcing and data
infrastructurecosts by
using a single platform <
20m
Derisking Compliance
by increasing rigor and
traceability of Financial
Crime data
7
Quick data analytics
in 2 to 4 weeks possible
through bundling relevant
data in one central place
Copyright © 2019Accenture.All rights reserved.
8. CASE STUDY: BRINGING ANALYTICS TO ACTION
8
New, machine learning approaches (random forest) are able to significantlyboost performance of detectingSARs
Machine learning model
Static data
Machine learning model
Static, account & TM data
SAR hit ratio: 66%
False positive ratio: 94%
SAR hit ratio: 91%
False positive ratio: 89%
Total alerts raised: 1,700 Total alerts raised: 300
Installed monitoring tool
Transactiondata
SAR hit ratio: 7%
False positive ratio: 99%
Total alerts raised: 2,500
Copyright © 2019Accenture.All rights reserved.
9. CASE STUDY: FUZZY NAME MATCHING
9
Additional / ad-hoc analyses can easilybe created,using the createddata infrastructureand analytics layer...
Fuzzy name process
Leak data sources Bank Data
Comparison
List on entry matches
Dictionary &
name harmonization
Dictionary &
name harmonization
Matching logic example1
Firestone LTD Fivestore LTD 2
Leaked data Client data Differentpositions
Matching results
Distance Client base
Perfect match >1002
1 Position 300
2 Postitions 2,000
• The developed crawler screened over 1 million leak data
entries
• Dictionary was setup and data sources were cleansed for
the screening process
1Levenshtein distance
The final list contained findings for further investigation on
potential fraudulent behaviour and the connection to financial
crime activities
2anonymized
“Cleaned”
leak data sources
“Cleaned”
Bank data
Copyright © 2019Accenture.All rights reserved.
11. WHY TRADITIONAL APPROACHES WILL FAIL
Difficulty deciding how and when to
source, progress and move new data
sources
Systems are not capable of supporting
new and evolving data sources
New capabilities are often siloed
efforts or POCs without clear vision
Difficulty building and managing
complex, hybrid data platforms
New self-service BI and analytics
models require an evolution of
traditional operating models
Considering talent pools is
increasingly important when making
tool and technology choices
11
Analytics ProcessesData
Copyright © 2019Accenture.All rights reserved.
12. IN OUR VIEW A DECOUPLED ECOSYSTEM IS AT THE HEART
OF THIS CHANGE
83% of C-suite professionals believe that their legacy systems hold value that they do not want to lose by changingto new
systems
12
TOO LONG (3-5 years)
TIME
CURRENTLY WE ARE HERE
Recognizing that legacy approaches are
too slow to adjust
A program to
fix things
EFFICIENCY
Cruising, focus
on new things
Slow
decline
Re-boot… (consolidate,
replace, multi-year
programs)
… and it continues, faster
and in shorter cycles
Copyright © 2019Accenture.All rights reserved.
13. 13
Streamline
Reduce high cost
workloads by
moving the data
out of the core
Build out new
functionality not
possible in the
core
Complement Hollow-Out
Replace
the core,
piece-by-piece
Replace
Replace
the core,
at once
5 components to decouple successfully
New
Build out
something that did
not exist before
GETTING THERE USUALLY REQUIRES
BANKS TO TAKE A STRUCTURED AND SEQUENCED
APPROACH
1 2 3 4 5
Copyright © 2019Accenture.All rights reserved.
14. … Modern Components
…
…
Coexisting Legacy as Systemof Record
Legacy in closed book run-off or decommissioned
.. … …
…
Channels
…
Greenfield Core
Real-time Data Lake Single Version of the Truth
Enterprise Back-End
Legacy Core
HR … Proc
… …
… … …
… … … …
.. …
.. … …
.. … …
.. … …
…
…
…
…
…
…
Pool 2
Data Access
Protection
MDM Pool 3 Pool n…Staging Area Pool 1
BCBS239 Reg Reporting
IFRS
Reporting
Posting
Engine
Enterprise
MDM
Doc Mgmt &
Archiving
Ops
Compliance
GL
…
…
…
…
…
……… … …
Write
APIs &
messages
Events Events Change Data Capture Streams
Digital DecouplingFilter relevant clients and dataTransform for consumption
StraightThroughProcessDIGITAL DECOUPLING IN ACTION
Event Streams
14
Decoupling Layer
Copyright © 2019Accenture.All rights reserved.
16. 16
Recommended approach
1. Focus on solving the data challenge, not the compliance challenge
2. De-couple your ecosystem / infrastructure
3. Consolidate relevant data sources
4. Setup integrated, shared analytics
HOW TO GET STARTED...
Copyright © 2019Accenture.All rights reserved.