ROI of Design: The analytics scorecard approach
The concept of virtual floorspace is that attention is valuable, and that e-commerce web sites need to evaluate how successful their designs are by the return they experience on each design component. I have a lot to say about analytics scorecards for tracking e-commerce design components after first covering research methods, but wanted to take an opportunity to introduce the topic by posting my response to George Papazian’s question in Linked In: “What are the best ways you have found to show Return on Design?”
The way Usography shows return on design is to create a metrics scorecard with very specific conversion goals, and then compare the scorecard results pre-launch and post-launch.
For e-commerce sites, the scorecard is simply based on revenue growth in categories relevant to the redesign. For example, did we sell more appliances with the new design, and if so, how many more? If the usability of a specific section was the topic of the redesign, e.g. checkout tunnel, then we’ll look at pathing and dropoff funnel pre-launch and post-launch.
In other types of sites, conversion is based on stakeholder goals that we capture in initial planning workshops, such as document downloads, time in a rich media piece, newsletter sign-ups, etc. A monetary value is placed on each conversion event, and then the scorecard results are compared with the cost of the redesign.
Previously we’ve used a six sigma assessment, but the assessment took a lot more time and energy to construct than analytics tools. Using Omniture, Coremetrics, Google Analytics, Webtrends, the data is easy to capture and compare apples to apples. However, one thing the six sigma study did better was measure the value of employee time, which may be completely missing from assessments based on agency design expenditures.
There are sometimes platform issues with tracking the same variables if the architecture has shifted dramatically from one release to the next. Another factor that complicates matters is when several aspects of the design change and you want to specifically attach a value or result to some design elements but not to others. Multivariate statistics are required to estimate the impact attributed to specific design changes.
Copyright 2009, Paul Bryan, Usography Corporation (www.usography.com)
Linked In: http://www.linkedin.com/in/uxexperts