CHRIS SMITH GETS FUNDED At a Deshpande networking session, a foundation director told us: "We don't care about your project, I mean it would be nice if you succeeded, but what we really care about is you. If this idea fails, we'll fund other one of your ideas in the spring." If you look at the Deshpande site: http://www.uml.edu/mvsandbox/default.html ...the quote makes sense. Deshpande builds student leadership, entrepreneurship, connections and confidence to develop the regional economy. A/B TESTING BETTER THAN GUESSING Deshpande is funding Chris Smith, a UMass Lowell Computer Science student, with $500 to set up a framework to measure MVHub usability. Deshpande's funding for Chris covers materials, not wages. In the next month, Chris will set up A/B testing scaffolding and as a separate project deploy http://mouseflow.com/ on MVHub and Northshoreportal.org . A/B testing is better than guessing. You slightly change the interface and see if that slight change improves a "conversion". A conversion is a measurable change in the user's actions. In the for-profit world, a typical conversion is a sale. If more people click the red "buy now" button than the blue "buy now" button, you know red buttons are better. Our goals are harder to measure. One of our draft conversions is: "clicks on a search result". Increasing clicks on a link called: 'Recovery, Alcohol Addiction “ might mean that we've helped more people get counseling. Increased clicks might mean that we've made the link more attractive without making it more relevant. “Helping people” is harder to measure than “increasing revenue”. By the way, Lara Swanson gave a great talk on A/B testing at the UML CS Dept Colloquium a couple weeks ago. Sean: http://thecsl.org/go/fake-blog/more-happy-users.shtml ... got us pointed to mouseflow this summer, but didn't have time to implement it. Mouseflow software will store little movies with exactly what visitors do with their mouse and keyboard each time they visit our sites, without capturing people's identities and raising privacy issues. We're looking forward to replacing guesses with data.
