9. Unpacking the Definition
• Efficiency = more outputs w/ less inputs
• Outputs = sales, leads, subscriptions,
engagement, etc.
• Higher efficiency = better Marketing ROI
• Often accomplished via A/B/n or MVT
testing
– Less ideal, but “change and observe”
counts, too
10. If Marketing is an Engine…
Image credit: Popular Hot Rodding
12. Optimization Philosophy
1. Be data-driven
– Data (quantitative & qualitative) drives testing,
not opinion
2. Be agile
– A small lift next week is better than a larger lift
next year
3. Lift + Learn
– It’s obviously about lift, but testing also provides
valuable customer/business insights
13. Optimization Philosophy (cont’d)
4. Gamble
– Some tests will lose
– Greater risk = greater potential lift
5. Baby steps
– Do what’s feasible ASAP, and build to more
complex tests
6. A process, not a project
– The long-term goal is testing as part of marketing
operations
15. What It Takes to Be Successful in Optimization
People
Tools Process
16. What It Takes to Be Successful in Optimization
People
17. People (Skill Sets)
• Business/Marketing Acumen
• Data Analysis
• Heuristic/Usability/UX
• Visual Design
• Optimization SME
• Copywriting
• Technical SME
• Project Management/Coordination
18. People (Skill Sets)
• Business/Marketing Acumen
• Data Analysis
• Heuristic/Usability/UX
• Visual Design
• Optimization SME
• Copywriting
• Technical SME
• Project Management/Coordination
19. What It Takes to Be Successful in Optimization
Process
20. Process (See? It’s Simple!)
Business Challenges Hypotheses
Analyze Report
Results Findings
Goals
Target Heuristic Design
Build Test
Next
Audience Data Variations
Steps
Usability Build
Web Data Design Test
Data Variations
21. Process (a simple one)
1. Define the problem
2. Research/observe the problem
3. Form a hypothesis
4. Conduct experiment
5. Analyze experiment results Look familiar?
6. Form a conclusion
7. Socialize results
22. Process (7 Simple Steps)
For more details:
http://brendan-regan.com/the-steps-of-the-
scientific-method-for-marketers/
Disclaimer: My personal blog’s snark does not represent the
views of Analytics Pros ;-)
23. Process (Different Models)
1. In House
– Build a team
2. Outsourced
– Leverage partners/vendors/agencies
3. Hybrid (most common)
– Outsource specialized expertise
• Data analysis
• Testing SME
• Creative
24. What It Takes to Be Successful in Optimization
Tools
25. Tools
• What technologies are needed to:
– Coordinate resources
– Design variations
– Document requirements
– Conduct experiment
– Analyze results data
– Socialize experiment results
26. Tools
• What technologies are needed to:
– Coordinate resources
– Design variations
– Document requirements
– Conduct experiment
– Analyze results data
– Socialize experiment results
28. Do These 4 Things Tomorrow
1. Conduct “gap analysis” on people
skills and tools
– Do we have enough resource
allocation?
– Do we need new tools?
– Do we have all the necessary skills in-
house?
29. Do These 4 Things Tomorrow
2. Get a testing tool in place
– Do we have IT support?
– Build vs. buy?
– Price range?
30. Do These 4 Things Tomorrow
3. Build a business case (if necessary)
– Case studies, calculators,
presentations?
– Opportunity cost?
– Tying to corporate goals?
– Who is scared of testing?
31. Do These 4 Things Tomorrow
4. Get resources and/or outside help
– Agency partners?
– Consultants/specialists?
– Testing vendors?
– Training?