What makes insights from Analytics more/less actionable? -not always billion dollar revenue generation. Slides walk you through the various components that make it actionable - challenges & what can be done about them. It was presented at Text Analytics Summit NY 2015.
1. Intended for Knowledge Sharing only
Actionability of Insights
Text Analytics Summit
Text Analytics Summit | June 2015
2. Intended for Knowledge Sharing only
Disclaimer:
Participation in this summit is purely on personal basis and not representing VISA in any form or
matter. The talk is based on learnings from work across industries and firms. Care has been taken to
ensure no proprietary or work related info of any firm is used in any material.
Director, Insights at Visa, Inc.
Help Executives/Product/Marketing with
actionable insights
RAMKUMAR RAVICHANDRAN
3. Intended for Knowledge Sharing only
Quick recap of what is it?
Quick recap of what it is
Intended for Knowledge Sharing only
What makes an insight actionable?
5. WHAT IS IT AFTER ALL??
Intended for Knowledge Sharing only
Specific answer to the question
Easy to understand
Timely & available (whenever, wherever & however needed)
Trustworthy & reliable
Scalable & Repeatable
6. Intended for Knowledge Sharing only
Quick recap of what is it?
Quick recap of what it is
Intended for Knowledge Sharing only
Seems easy enough?
7. SEEMS EASY, SO WHERE IS THE PROBLEM?
Intended for Knowledge Sharing only
Sky high expectations (Data Scientists the new “Alchemists”)
Philosophical differences
Vague Questions/Undefined
Constraints (Data, Time, Resources unavailable)
Heartbreak Syndrome “not what I expected you know, answer”
9. 9
CUSTOMER EDUCATION ON MULTIPLE VALUE PROP WOULD HELP
Intended for Knowledge Sharing only
Size behaviors
with KPIs and
high level
drilldowns
(Sizing)
Inform Investigate Predict Optimize Mine
Root cause
analysis:
Hypotheses
testing via data
drilldowns
(Business
Analytics)
Determine
Causal
relationships
(Advanced
Analytics)
Experiments on
options to
verify which
one works
(A/B Testing)
Automated
relationship
discovery and
Data Products
(Machine
Learning)
10. 10
SEEMS EASY, SO WHERE IS THE PROBLEM?
Intended for Knowledge Sharing only
Sky high expectations (Data Scientists the new “Alchemists”)
Philosophical differences
Vague Questions/Undefined
Constraints (Data, Time, Resources unavailable)
Heartbreak Syndrome “not what I expected you know, answer”
11. PHILOSOPHICAL? WHAT DO YA MEAN?
Intended for Knowledge Sharing only
"Analysts are from Mars and their customers are from Venus…
ANALYSTS CUSTOMERS
Language they speak
Numbers, graphs, lifts,
accuracy, insights
If we do this, then we get
this...
12. GRAPHS, OH SO MANY OF THEM!
Intended for Knowledge Sharing only
14. PHILOSOPHICAL? WHAT DO YA MEAN?
Intended for Knowledge Sharing only
Analysts are from Mars and their customers are from Venus…
ANALYSTS CUSTOMERS
Language they speak
Numbers, graphs, lifts,
accuracy, insights
If we do this, then we get
this...
What excites them Brilliance of approach Simplicity of the answer
15. THE SEARCH FOR TRUTH…
Intended for Knowledge Sharing only
16. WHEN ALL THEY WANT IS…
Intended for Knowledge Sharing only
17. PHILOSOPHICAL? WHAT DO YA MEAN?
Intended for Knowledge Sharing only
Analysts are from Mars and their customers are from Venus…
ANALYSTS CUSTOMERS
Language they speak
Numbers, graphs, lifts,
accuracy, insights
If we do this, then we get
this...
What excites them Brilliance of approach Simplicity of the answer
How they think Detail oriented Big picture
18. HMMM, WHAT IF WE DO THIS?
Intended for Knowledge Sharing only
20. PHILOSOPHICAL? WHAT DO YA MEAN?
Intended for Knowledge Sharing only
Analysts are from Mars and their customers are from Venus…
ANALYSTS CUSTOMERS
Language they speak
Numbers, graphs, lifts,
accuracy, insights
If we do this, then we get
this...
What excites them Brilliance of approach Simplicity of the answer
How they think Detail oriented Big picture
What they can
compromise on
Time for accuracy Perfection for timely action
22. BUT SHE IS LIKE
Intended for Knowledge Sharing only
23. PHILOSOPHICAL? WHAT DO YA MEAN?
Intended for Knowledge Sharing only
Analysts are from Mars and their customers are from Venus…
ANALYSTS CUSTOMERS
Language they speak
Numbers, graphs, lifts,
accuracy, insights
If we do this, then we get
this...
What excites them Brilliance of approach Simplicity of the answer
How they think Detail oriented Big picture
What they can
compromise on
Time for accuracy Perfection for timely action
Biggest difference Scientists who deal with facts Artists who deal with gut
25. SEEMS EASY, SO WHERE IS THE PROBLEM?
Intended for Knowledge Sharing only
Sky high expectations (Data Scientists the new “Alchemists”)
Philosophical differences
Vague Questions/Undefined
Constraints (Data, Time, Resources unavailable)
Heartbreak Syndrome “not what I expected you know, answer”
27. SEEMS EASY, SO WHERE IS THE PROBLEM?
Intended for Knowledge Sharing only
Sky high expectations (Data Scientists the new “Alchemists”)
Philosophical differences
Vague Questions/Undefined
Constraints (Data, Time, Resources unavailable)
Heartbreak Syndrome “not what I expected you know, answer”
28. HOW TO MANAGE CONSTRAINTS
Intended for Knowledge Sharing only
•Tactical Prioritization: Classify the requests into “Firefights, Urgent and
important, Important but not urgent, Good to have” based on factors like
Requestor, Urgency, Impact and availability of resources solve it.
•Pre-Analysis work: Strategic prioritization (Outcome Focused), Gap analysis on
data/proxy, Various approaches and the Sizing of ETA for each, final output
templates.
•Expectations setting: Discuss with requestors, the output from Pre-Analysis and
decide together on next steps. Set up Milestones/regular check-ins.
•Execution, Communication, Fine tuning & Course-correction (if necessary)
•Automate if necessary
29. STRATEGIC PRIORITIZATION (ILLUSTRATIVE)
Intended for Knowledge Sharing only
Sl.
No.
Ask Why is it needed?
How will it
be used?
Fit with the high level
business Strategy/KPI
impacted
Plan B?
1
What are the
Consumers
saying?
Redesigned the
website and need
to know the
customer reaction
Roll back/
Ramp up
based on
feedback
Customer Satisfaction
& Engagement via
better website UX
A/B Test
findings only
2
Thematic
Extraction of
Tweets
Investigate why
Site Engagement
(#Page Views/Visit)
down WoW
Feedback will
inform where
issues were
and need to
be addressed
Maintain Product
uptime for the users
Pathing
Analysis/
Heatmap
Analysis only
30. OUTPUT CUSTOMIZED TO CONTEXT OF THE USER (NEED, TIME, MINDSET)
Intended for Knowledge Sharing only
“In mail”
Recommendations
with supporting
graphs, tables, etc.
“Story Deck”
Full deck with the pitch
and supporting arguments,
numbers, graphs, charts
“On-the-go”
-Mobile App, On the
Cloud, Subscriptions
-Reports, Dashboards,
Infographics
Algorithm/Model
Ready to be deployed
How to decide? Customer needs;
Turnaround Speed; One time/reuse;
Deployment on Front end; Strategic
Doc; Quick read/research doc
31. CUSTOMER DRIVEN ANALYTICS
Intended for Knowledge Sharing only
Pre-work & Kickoff1 Analyst, Customer
Translation to Analytical
Framework
2
Analyst, Researcher, Data Instrumentation, & Data Manager,
Developer, Data Scientist
Data Collection and
Preparation
3 Analyst, Data Manager, Data Scientist
Analysis, Validation &
Verification
4 Analyst, Data Scientist, Customer and SME, Researcher
Actionable insights and
impact sizing
5 Analyst, Customer, Leader
A/B Testing6 Analyst, A/B Testing, Customer, Developer
Rollouts7 Customer, Leadership & Executives
ResponsibleSteps
32. SEEMS EASY, SO WHERE IS THE PROBLEM?
Intended for Knowledge Sharing only
Sky high expectations (Data Scientists the new “Alchemists”)
Philosophical differences
Vague Questions/Undefined
Constraints (Data, Time, Resources unavailable)
Heartbreak Syndrome “not what I expected you know, answer”
33. COME ON BOY, DO IT NO…
Intended for Knowledge Sharing only
35. FINALLY SOMETHING ON TEXT ANALYTICS
Intended for Knowledge Sharing only
Complex and takes time to execute(Data collection/standardization/
cleaning up/Preparation too long the first time)
Chance of Model/Analysis not revealing anything (junk data)
Chance of Model/Analysis giving false positives (since sparse data issue)
Low RoI Exercise (too much effort for little incremental benefit)
Dependency on others – researchers, instrumentation, etc.
Specific Risks with Text Analytics…
Rigorous pre Analytics assessment and expectations setting.
Success Criteria to be changed from Significance to Consistence/raw
counts.
Impact Sizing and Vetting the impact with A/B Testing.
…so what can be done about it
36. Intended for Knowledge Sharing only
Quick recap of what is it?
Quick recap of what it is
Intended for Knowledge Sharing only
Wrapping it up
37. WHY IS THE TOPIC SO POPULAR NOW?
Evolution in the value prop of Analysts:
What/where/how much -> what can happen ->what should we do ?
Audience has broadened (A numbers middle man -> Front line Managers)
Luxury of time has evaporated
Nature of questions have drastically changed (Expectation of being able to
connect the dots in “Data Lake” world).
Overselling potential before getting “there”
38. REALLY WRAPPING IT UP, I PROMISE…
• “Know” that not all Analytics is supposed to be actionable.
• “Must have” User Experience Design (UED) Strategist for the Analytics
practice
• “Ensure” Deeper Stakeholder involvement in Analytics development & Test
& Learn approach must
• “Develop” Outcome Focused Approach for Analytics
• “Prepare” for ever more increasing ask for analytics and related
actionability issues
Putting it all together…
39. Intended for Knowledge Sharing only
Quick recap of what is it?
Quick recap of what it is
Intended for Knowledge Sharing only
Appendix
40. Intended for Knowledge Sharing
only 4040
THANK YOU!
Intended for Knowledge Sharing only
Would love to hear from you on any of the following forums…
https://twitter.com/decisions_2_0
http://www.slideshare.net/RamkumarRavichandran
https://www.youtube.com/channel/UCODSVC0WQws607clv0k8mQA/videos
http://www.odbms.org/2015/01/ramkumar-ravichandran-visa/
https://www.linkedin.com/pub/ramkumar-ravichandran/10/545/67a