how the economist with cloud bi and looker have improved data-driven decision making
TRANSCRIPT
PROOF OF CONCEPT IN WEEKS,DATA SOLUTION IN MONTHSImproving data-driven decision making
BobbyGill
PeteGrant
Sebastien Fabri
Where it started
Objectives
1. Provide a platform that allows future scalability, flexibility and agility to consume new customer data inputs.
2. Progressively consolidate customer data silos into a single customer view repository based on value.
3. Provide a platform that will allow more sophisticated analysis (e.g. predictive modelling).
4. Provide the business with self-service capability to key metrics and dashboards through the use of a data analytics and visualisation tool.
Platform Size
Implementation Approach1.Available platform and software services over
development and apparent control• Time is the most scarce resource• We’re not set up to COMPETE with AWS…
Services Without Commitment
More effective to use some of this…
Services Without Commitment…than to make our own from scratch
What We Actually Did Part 1• We used S3, Data Pipeline, EMR, Redshift -
the solution is AWS specific• We used Lambda which is AWS specific in that
it calls AWS services• We used CodeCommit• We used Jenkins as a scheduler
Implementation Approach1.Available platform and software services over
development and apparent control2.Creative assembly of services over product
customisation• Any one product might not satisfy all requirements• Customising products might not be the best option
Custom AssemblyCustom is better
Customising the product might work
But the gap isn’t usually just a paint job
What We Actually Did Part 2
Implementation Approach1.Available platform and software services over development and apparent control
2.Creative assembly of services over product customisation
3.Empowering the end user over charging for effort
• Avoid writing and reading specs• Cut out the middle men
What We Actually Did Part 3
GOVERNANCE
DATA
UI & VIZ
TRIGGEREMAIL
STREAM
API
Where Do We Start?
UI & VIZ
TRIGGEREMAIL
STREAM
API
MODELING LAYER(LookML)MP
P
How Do We Deliver It?
REDSHIFT DATA TEAM DATA PLATFORM END USERSINNOVATION
THE ECONOMIST
LOOKER
The Economist Implementation
REDUCED DATA PREPARATION AGILE DEV TRUST THE DATA EMPOWER
USERSNEW INSIGHTS
INNOVATION
THE ECONOMIST
LOOKER
The Economist Value
Time Lines
PoC.
Month 1 Month 2 Month 3 Month 4 Month 5-6
CEO requests for combined Data Vision
Internal reviews and requirements gatherings
PoC Built and tested!AWS Environment and Looker Dashboards
Business Review Road show
Building Business Case
Production
Month 8-9 Month 10 Month 11 Month 12
Production System Build - AWS
Internal Agile workshops - Identify the top 5 deliverables
Looker Dashboard build-Stick Rate reporting-Customer Journey
Phase 1 sign off and Phase 2 planning
Feedback
1. Our project lets us tie data together, leveraging data across the whole group, in order to spot practical, actionable insights. This information is crucial to our business.
2. The potential to drive real insight becomes suddenly considerably easier.3. We are now armed with facts and can set out priorities, i.e. to develop
features to our app and site to spur usage. 4. Cross-platform analysis, for the first time, we can begin to understand
the impact of registration and digital usage on conversion and retention
Questions?