Master Conversion Optimization
- 2. Conversion Optimization
• What is the goal of conversion optimization?
• Growth
• How do we optimize the website so that the business will grow?
• SEO is clogged and difficult
- 3. It’s About Marketing
• Figure out which pages cause the biggest drop-offs
• Can understand your customers better
• Finding where the problem lies, you can identify the what
• Most of the world still operates by guessing what might be a good change
and hope sales will go up
• They don’t use the data
- 4. Use Data to Form Hypotheses
• Turn unsupported and baseless opinions into data-informed, educated
hypotheses
• Know what’s happening and understand why
• Set goals, set up measurement, gather data, analyze data, turn data into an insight, test
hypotheses, get data from tests, go back to data analysis. Repeat. Repeat. Repeat.
- 5. The Data-Informed Approach
• We need better data
• Gathering only “must have” data
• Avoid analysis paralysis
• Two steps/frameworks are extremely important:
• Technical analysis
• Heuristic analysis
- 8. Expanding Technical Analysis
• Bugs are the main conversion killer
• Fixing annoying or functionality errors amongst different platforms/interfaces will be a
huge improvement
• Speed analysis
• Page load and interactive time to ensure the user doesn’t have to wait for pages to load
• Use Google PageSpeed Insights to list out problems, then resolve these issues
• Ping website to understand how long it takes to relay information back and
forth from the server to an end-user
- 9. Expanding Heuristic Analysis
• Almost like sharing an opinion, but not quite
• Can be done quickly
• Optimizing using results-based
experience
• Avoid random comments
• Best done in a group
• Use screenshots to capture ideas, comments, etc.
- 10. Become a Google Analytics Expert
• If you’re skilled, as most marketers aren’t, time to learn is a must
• Career depends on it
• Need to know impact of ideas, proven through data
• Know in advance what you want to analyze – figure out the “what”
• Use qualitative research and heuristic analyses are the best for figuring out the “why”
- 11. Measure Important Stuff
• You’re ahead of most companies if you only measure what matters
• KPI – key performance indicator
• Helps you understand how you are doing against objectives
• Ex. Conversion rate, revenue per visitor
• If you can’t figure out why something is important,
- 12. Mouse Tracking and Heatmaps
• Heat Maps
• Representation of mouse action or hovering on page
• Red is hot, blue is cold
• Click Maps
• Representation of where people have clicked
• Make words or place people want to click, without a link, into one
- 13. Scroll Map
• Sees how far a user typically scrolls down a page
• Helps to understand where to put priority information
• Helps you decide where you need to tweak designs
• Can be used to understand what users think is the “must have” information
• Watch scroll maps and see where people stop scrolling
• Add things like color lines to tell people something is different below
- 14. Website Surveys
• On-page Survey – someone fills out a survey while on your site
• Both can be useful
• Help to derive important data
• To ensure that people don’t get frustrated, only do surveys for a short time
• Only ask for important, measurable data – help improve KPIs
• Exit Survey – as someone is about to leave, you hit them with a popup
- 16. User Testing
• Creating protocols
• Limit one session to 20 minutes
• Ensure key actions are understandable
• Identify friction
• Use test user mistakes as a way to see
what of your website sucks
• 5-10 testers is enough
- 17. Data Collection to Test Hypotheses
• Placing problems and traffic leaks into 5 buckets for translation and solution:
• Test
• For traffic leaks, you test different hypotheses to see what is going wrong
• Hypothesize
• Found problematic page, but not sure where the problem is
• Instrument
• Reporting in analytics is poor and needs to be re-evaluated
- 18. And These Buckets…
• Just Do It
• Fix is easy to identify and it just needs to get done; Do it!
• Investigate
• Some testing needs to be done with particular devices to triangulate the problem
- 19. A/B Testing: : How to Get it Right
• It’s time to test the hypotheses
• Pick appropriate testing tools for each hypothesis
• Get a developer to help
• Sample size needs to be big enough
• In order to test validity of results, a valid sample size is needed
• Can use statistical calculators; all available for free online
- 20. A/B Testing, Cont’d
• The more segments tested, the higher sample size needed
• Test has to be long enough
• Just because the sample size is valid, enough
time must be allowed for a full business cycle
to get a good representation of data
- 21. Learning From Your Results
• First off, be prepared to test.
• A lot. Then some more. Then again.
• Inconclusive tests happen often and are okay
• Usually happens with a wrong hypothesis
• Learn from failure