Title: Data : Considering more than the obvious Presented at both Leading the Product 2017 - Melbourne and Sydney

1. Leading The Product 2017
Speaker Slides
Melbourne and Sydney, Australia
Wendy Glasgow
Google
For more information go to
www.leadingtheproduct.com

2. Confidential + ProprietaryConfidential + Proprietary
Data: Considering more than the obvious
wendyg@google.com

3. Confidential + Proprietary
Data
to define
what we
build
Data
generated
from our
product
Data
powering
our
product
2 31

4. Confidential + Proprietary

5. Confidential + Proprietary
Data
to define what
we build
1

6. Confidential + Proprietary
Metrics to define success - moving away from the gut
Get them right - we can truly build a great product that grows our business
Get them wrong - we can look successful on paper but completely miss the mark

7. Google Confidential and Proprietary
weight target quantity target
Russian nail factory workers
Early 20th century

8. Macro:
What is my business trying to do?
What is my team trying to do?

9. Build a strong profitable business

10. Proprietary + Confidential
Executives that adhere to metrics that tie directly to
business objectives
3xmore likely to
hit their goals
Source: New Study Reveals Why Integrated Marketing Analytics are Critical to Success, Think with Google, Forrester, March 2016

11. Confidential + Proprietary
Understand the macro before diving into the micro
Product Strategy
Market Penetration | Brand Positioning | Profit Targets
Product Plan
Metrics
Marketing Str… Sales
Cost / Service External DriversEngagement
Product
Teams
Your Product
strategy
enables focus

12. Confidential + ProprietaryConfidential and Proprietary
Jack Welch, Former CEO of GE
“There are only two
sources of competitive
advantage:
The ability to learn more
about our customers
faster than the
competition,
and the ability to turn that
learning into action faster
than the competition.”

13. Proprietary + Confidential
Customers are NOT created equal
Focus: Who is your customer?

14. Confidential + Proprietary
Know your ‘best’ customers
$ $$$$$
$$$$$
$
Value Spend less
Cost to Acquire
Ideal
Customer
Base
Cost to
Service,
Support,
Retain

15. Confidential + Proprietary
Widen the scope of considered data
Marketing
Ad Logs
Search
DM EDM
SMS
Competitions
Newsletters
Product,
Websites
Stores
Engagement
Analytics
Transactions
Customer
Services &
Support
Call centre
Customer
interactions
Finance
Transaction
Business
Costs
Operations
& Logistics
Operational
costs
Delivery
Sales
CRM
POS
Customer
Value
People &
Culture
People costs
Skills matrix
Attrition
Tech
POS
Logs
Analytics
Relevant Data
Data Strategy
Data Governance

16. Confidential + Proprietary
Data generated
from the product
2

17. Confidential + Proprietary
● Deal with data early
● Ensure you have a data strategy
● Add a section at the definition stage
● Make it mandatory
● Provide a process, make it consistent
● Over capture and
LABEL!
You’re focused on getting your product live

18. Confidential + Proprietary
We’re capturing data - what is important to consider?
● Data storage is CHEAP - $0.02 per TB per month in BigQuery!
● Capture everything possible - but make it readable
● Consider the data points you’re capturing
● Make the data meaningful - LABEL
● Link your data - IDs, Labels
● Timestamps are critical

19. Confidential + Proprietary
Tools
BigQueryAttribution
Data Studio
Data Vis
Tag Manager
Tag Mgt
Optimize
Testing/Personalisation
Google Analytics
Analytics

20. Confidential + Proprietary

21. Confidential + Proprietary
Make friends
with statistics
- OR -
with someone
who already is
correlation
causation
Get intimate with your data
relevance

22. Confidential + Proprietary
Data as an Asset

23. Confidential + Proprietary
There is no competitive advantage within an organisation!
Share
your
Data
There is no competitive advantage within an organisation!

24. Confidential + Proprietary
Data
powering our
product
3

25. Confidential + Proprietary
Machine Learning is the new
ground for gaining competitive
edge & creating business value
*Source: MIT Survey 2017; n=375
Bain Consulting Study
Competitive advantage ranked as top goal
of machine-learning projects for 46% of IT
leaders & 50% of adopters can quantify ROI
2X more
data-driven
decisions
5X faster
decisions
than others
3X faster
execution

26. Confidential + Proprietary
Machine Learning Allows You to Solve a Problem Without
Codifying the Solution
✓ Recognizes patterns in data
✓ Predictive analytics at scale
✓ Builds ML models seamlessly
✓ Fully managed service
✓ Deep Learning capabilities
Google Cloud AI

27. Confidential + Proprietary
First Step in This Journey Begins with
Data
“Every Company will be a Data Company”
*Source: Wired, Bloomberg, Fortune, McKinsey
Proprietary + Confidential

28. Confidential + Proprietary
Machine Learning Lifecycle at a Glance
How do I collect, store and
make data available to the
right systems?
How do I understand what data
is required to solve my business
problem?
User
Data Objective
TrainServe
How do I get to a working
model within the period of
time where my objective is
still relevant?
How do I scale prediction
into production systems?
How do I keep my model
relevant with continuously
updated data?

29. Confidential + Proprietary
Flow to build a custom ML model
Identify
business
problem
Develop
hypothesis
Acquire +
explore data
Build a
model
Train the
model
Apply and
scale
1 2 3 4 5 6

30. Confidential + Proprietary
Structured Data
● Spreadsheets, Logs, Databases
● Text that includes structure
● Data needs to be separated
● Typical data generated from products
Unstructured Data
● Natural Language, Images
● More complex but sometimes these are
better understood
● Number of existing ML APIs -
Supervised Learning
● Need labels on the data
● Build examples to train the system
Unsupervised Learning
● Data is grouped / clustered
● Drawing inferences from data sets

31. Confidential + Proprietary
A feature in ML is very different from a feature in Product
In ML, a feature is an individual measurable property or
characteristic of a phenomenon being observed.
Choosing informative, discriminating and independent
features is a crucial step for effective algorithms in
pattern recognition, classification and regression.
Feature Engineering

32. Confidential + Proprietary
A feature is a data point, so what is good?
Represent raw data in a form conducive for ML
1. Should be related to the objective
2. Should be known at production-time
3. Has to be numeric with meaningful magnitude
4. Has enough examples (absolute minimum of 5)

33. Confidential + Proprietary
What can I do today to plan for ML
1. Find your Data Strategy and Governance owners –
get familiar with it or create it!
2. Identify the decisions your product makes today.
3. Consider suitability for automation with ML.
4. What data do you have today and what do you need
to capture?
5. Capture data in line with your strategy and
governance guidelines – update them if necessary.
6. Capture LOTS of data, but LABEL it well and
consistently!

34. Confidential + Proprietary
Takeaway 1 Takeaway 2
Value is in
use of data
Think
inside, outside
& future
It’s what we do with the data that matters
BUT… early consideration can increase value
How does you relate to your surroundings
Relevance, correlation and causation

35. Confidential + Proprietary
● Predictive maintenance or
condition monitoring
● Warranty reserve estimation
● Propensity to buy
● Demand forecasting
● Process optimization
● Telematics
Manufacturing
● Predictive inventory planning
● Recommendation engines
● Upsell and cross-channel
marketing
● Market segmentation and
targeting
● Customer ROI and lifetime value
Retail
● Alerts and diagnostics from
real-time patient data
● Disease identification and risk
satisfaction
● Patient triage optimization
● Proactive health management
● Healthcare provider sentiment
analysis
Healthcare and Life Sciences
● Aircraft scheduling
● Dynamic pricing
● Social media – consumer
feedback and interaction
analysis
● Customer complaint resolution
● Traffic patterns and
congestion management
Travel and Hospitality
● Risk analytics and regulation
● Customer Segmentation
● Cross-selling and up-selling
● Sales and marketing
campaign management
● Credit worthiness evaluation
Financial Services
● Power usage analytics
● Seismic data processing
● Carbon emissions and trading
● Customer-specific pricing
● Smart grid management
● Energy demand and supply
optimization
Energy, Feedstock and Utilities
Cloud Machine Learning Use Cases