2. Agenda
● GO-JEK’s Business Intelligence at a glance
● Big Data?
○ Use Case
○ Big Data in GO-JEK
● Design Rationale
○ Audience of Data
○ Data Hierarchy
○ BI Architecture
● Current Stats
● Takeaway
Indonesia Developer Summit 2017
5. GO-JEK’s Business Intelligence at a glance
● One of three independent data teams, other than Data Engineering (focusing on data pipeline in
the entire company) and Data Science (focusing on AI and future data product).
● Top partner of Product Owners, consultant of every products in GO-JEK, delivers insight and
proves hypothesis based on data.
● Distributing business data throughout the company: makes data available, aligned with
business and accessible by business users.
Indonesia Developer Summit 2017
7. Use Case
“Can you help to analyze customer behaviour in our
apps when they book a GO-RIDE for the last one
month?”
Indonesia Developer Summit 2017
8. Use Case - cont’d
Indonesia Developer Summit 2017
Select Service Set Pickup
Set
Destination
Place Order Waiting Order
Order
Completed
9. Use Case - cont’d
Indonesia Developer Summit 2017
● Input
○ ~150 mio events in daily basis
○ Export data from API
○ Heavy analytic processing is
needed
● Output
○ Funnel analysis
○ Insight and Recommendation
10. We can have more questions...
“... Can you do it for another GO-JEK service?”
“... Does the OS of customer device have a
correlation with this?”
“... Can we have the details of those customers
in certain steps?”
“... Can you divide it based on app version?”
“... How many from them is our high value
customer?”
Use Case - cont’d
Indonesia Developer Summit 2017
Or other use cases…
“... Can we do location analysis where our driver
usually stops?” → a lot more of data! e.g. GPS
ping (approx. 850 mio in a day)
“... We need to know more about our customer’s
feedback” → more data is better; e.g. NLP
“... I want to do statistical modelling from time of
each event during the flow of order. Can you
help us?” → heavy calculation and processing
12. What we learn from our use cases?
● Data is — just data, no matter what the
state is.
● The process (most of the cases) is not
easy enough to be done by everyone —
creating the needs of specific skillsets to
gather the insight.
● Everyone is curious with everything and
want to do it on their own (if it’s easy
enough)
All is part of our culture as a data-driven
company
Big Data in GO-JEK
Indonesia Developer Summit 2017
What did we do?
14. Who get access to
data?
High Level Management
Business Analyst
(and everyone else)
Data Analyst
What kind of data
granularity ?
Low — business summary
data
Medium — combination of
summary and raw data
High — raw data
Why they need it? Decision Support System Hypothesis Analysis
Exploratory Analysis, System
Creation
How they access it?
Worksheets or reports with
simple slice and dice
operations
Worksheets or reports with
simple slice and dice
operations
SQL query for complex
analysis
SQL query for more complex
analysis
Custom code for statistical
analysis, visualization,
machine-based decision
making, etc.
Audience of Data
Indonesia Developer Summit 2017
15. Data Hierarchy
Indonesia Developer Summit 2017
Data Lake
Data
Warehouse
Data Mart
Low
High
Granularity
Variety
Availability
High
(High-level
Management)
High
(Data Analyst)
Access
LevelLow
(Business
Analyst)
High
Low
Governance
16. BI Architecture
Indonesia Developer Summit 2017
What do we want?
● High performance, scalable, minimum
operation maintenance
● Full resolution dataset
● Easy data discovery process
What do we built?
● Leveraging available cloud features, e.g.
BigQuery
● Data modelling, creation of denormalized
tables
● Easy to use front-end, e.g. Metabase
● Creation of universal data dictionary, e.g.
Datadex
18. Indonesia Developer Summit 2017
> 30%
*This is only business metrics data
collected by BI.
Growing Data Volume per Month
> 150
*Contains full resolution, reusable summary
and roll-up dataset.
Multi Resolution Dataset
> 6000
Metabase Cards & Tableau
Dashboards
~ 500
Average Daily Metabase & Tableau
Users
*Everyone just loves data!
Current Stats
> 1000
Data Points
From over 35 internal and external data
source
19. ● Built your data solution around business, and keep it relevant every time. Data is built
and comes from business activity, so make sure it goes back and help the business itself.
● Storing the data is important, so is the presentation of it. There is no use if the data is
stored well but not easy to be accessed by everyone.
● Every questions will lead to business decisions. Make sure every questions and
hypothesis can be answered and proven with data as easy as possible.
Takeaway
Indonesia Developer Summit 2017
20. We believe in creating infinite opportunities to unleash social
impact through technology.
Do you share the same belief?
Indonesia Developer Summit 2017
bit.ly/GO-JEKDEVSUMMIT