Upload
charissa-reyes
View
28
Download
0
Embed Size (px)
DESCRIPTION
Ammunition for the CFO: How to Be (or Satisfy) a Hardnosed Customer for Analytics. JC Herz [email protected] February 28, 2012. When You’re at Third Base…Get Tested!. DATA AUDIT is a gating condition Quantity Sometimes it’s not as big as you think it is Speed Number of sources Quality - PowerPoint PPT Presentation
Citation preview
Ammunition for the CFO: How to Be (or Satisfy) a Hardnosed
Customer for Analytics
JC Herz
February 28, 2012
When You’re at Third Base…Get Tested!
• DATA AUDIT is a gating condition– Quantity
• Sometimes it’s not as big as you think it is• Speed• Number of sources
– Quality– Who “owns”
Resources
• $$: None, Some, Lots, Obscene Amounts
• People: How many, how good, who owns
• Bandwidth: Organizational attention
• Transition points
• Time: have to show results when?– A week? A month? Next quarter?
OODA Loop and its Pitfalls
• Observe, Orient, Decide, Act - where does this project play?
• Observe: – Outward-facing lamp post problem (social media)– Internal: lift the rug: system is being
• a) Measured• b) Managed• c) Gamed• d) Gamed with the complicity of managers
OODA Pitfalls: Orientation
• Misunderstanding of: resources, identity, mission, market position
OODA Pitfalls
• Decide: Is leadership undertaking a Big Data analytics effort as a way to avoid making decisions?
• Act: cultural and political fallout, internally and externally: data driven decisions have consequences
Locus of Effort
• Internal team - dedicated• Croudsourced inside organization
– IT + subject matter experts: the odd couple
• Consultants • Vendor• “How does this engagement build capacity
within my organization”– Pretend you’re the benign dictator of a developing
country, across the table from an oil company
Vendors
• What does the vendor really sell?– Consulting services?
• Bodies, Expertise, Coming up with a better algorithm
– Software licenses?– Infrastructure: bandwidth, storage, CPUs– Everything looks like a nail
• The iron triangle– In construction: time, cost, quality– In analytics: storage, cycles, performance– Demand three cost scenarios that minimize each