Insurance comparison engine Information Management ProjectSpeaker: Miloud Belkacem
24 March 2016
Agenda
Project Presentation
B&D Answer
Benefits and Conclusion
2
301/05/2023
Amongst the market’s top 4
platforms
Activity exclusively web
oriented
Significant monthly web traffic
Insurance comparison platform active on the French Market
Who is the Client ?
Startup
401/05/2023
Client’s Business Model
Site visitors fill-in forms to compare insurancesThe company sells the visitors’ forms to partnersPerformance of the organisation relies on Website Traffic Conversion Rate
501/05/2023
Project Objectives
Business & Decision Belgium was selected to define & execute the client’s data strategy
Two parallel tracks: Big Data & Analytics and Business Intelligence
The key high level objectives being:
Competitive Edge
Helping the client gain competitive edge and foster its
market position
Boost Insights
Leverage analytics practices to better understand what
happened before, what is happening now and what could
happen in the future
Modernize the Data Platform
Set up a modern data lab based on top-notch technological
solutions and powerful practices
6
Initial Situation
01/05/2023
Platform
Site DatabaseWeb Logs
Data Scientist
Studio Excel file
Management Line & decision takers
Deliver Results
IT Department
Relay Decisions
Implement
Large volumes of Raw dataSlow and heavy analysis
Limited insights and analysis capabilitiesSlow cycle to market Ads and Targeting
Slow adaptations of the model
ExtractGenerate
Analyze
701/05/2023
Challenges
Difficulty to exploit large sized web logs which is key to understanding the behavior of users
Tedious manual data extraction to perform analysis due to performance and the need to perform data transformations
Slow Analysis Life-Cycle as: Data Scientist delivers information manually and irregularly to business Decision takers assess the analysis results and take decisions IT builds new recommendation Ads and targeting rules into the platform based on the input
of the management which creates latency
Agenda
Project Presentation
B&D Answer
Benefits and Conclusion
8
901/05/2023
B&D’s Mission
Query & Analysis Solutions03Self-Service
Limited BI / Spreadsheet05Limited
Strategic02 Dashboards management
Operational04 Operational Reporting
Data Mining & Predictive Analysis01Analytics
Excellence
Client
Approach Overview
In order to overcome the challenges detailed earlier, B&D has:
Insurance Comparison Platform
Business Intelligence
Big Data & Analytics
Selected Microsoft as the technology provider Set up a full-featured Data platform hosted on Microsoft Azure Define data governance to streamline reporting efforts
Design a BI solution to Deliver traditional BI outputs (reports, Ad-Hoc, etc.) Serve as the destination of aggregated Big Data
Set up a data lab on the cloud to Load and make available large sized web logs and external files Provide data scientist tools for analysis purposes Deploy Machine Learning platform and mechanisms Plug & Play
10
1101/05/2023
Azure Machine Learning in a Nutshell
Machine Learning cloud based component
Provides trained & enriched predictive models
Provides web service based interface to integrate with third party tools Implement Real-Time targeting and Ads selection Real time suggestions Automatic referrals Churn calculations Customer segmentations Next best offer ...
Azure ML
1201/05/2023
Empowered Insight platform
Insurance Comparison Platform
Site Database
Web Logs
Data Scientist
Studio
BI Load
Generate
Business Intelligence
Load
Large volumes storage
Machine Learning
Consume
Analyze
Data Lab
Automated Real-Time Targeting and Ads selection
HDInsight
Cons
ume
1301/05/2023
Single Data Platform
In-HouseSources
ConsumptionPlatformData HubStaging area
ML StudioWebLogs
Data retrieval
ReferenceFiles
Manual Cnsolidation
External Sources
Insurance Files
Extract Transform Load Consolidate
Mirror
LZ MERStaging
BigData Stage(Hive metastore)
Transform M
erge Load
Cube
Process
MER
Direct Access
1401/05/2023
Zoom on Azure ML
Agenda
Project Presentation
B&D Answer
Benefits and Conclusion
15
FUNCTIONAL
Benefits of the solution
16
More relevant recommendations
On-time recommendations
New requests for contact (MER)
Increase conversion rate
1
2
3
4
FUNCTIONAL TECHNICAL
Improved data integration (data flow)
Fully automated recommendation system
Usage of state-of-the-art ML technology
Usage of the cloud infrastructure
SocialAnalytics-Ready
Benefits of the solution
16
More relevant recommendations
On-time recommendations
New requests for contact (MER)
Increase conversion rate
1
2
3
4
Conclusions
Combine traditional BI & Big Data capabilities
Project hosted in the cloud
Project initiated overnight & first results presented after a few weeks
« Data Lab » solution to validate use cases, then industrialization
Machine Learning capabilities activated
BIBigData
17