Yes
Good points, Noemi. Re 2, I added links here. In general, we could add links that go from the minisite into the workspace.
“Count me in”… the hard part is building alternative pages to test! I think we would use our time well in testing:
- the home page, obviously. About 12% of all page views...
- the profile page. This is less obvious, but then I remembered reading of this pattern from Ning. On Edgeryders, more than 8% of all page views since we are tracking are profile pages. Our profile pages are ugly as hell, with too many fields and some missing information! Incidentally, I know how many profile pages get viewed because there is a neat trick to count classes of pages dynamically generated: go to the Content => All pages report in GA. Look for the search field, then click on "advanced" and select a search "Starting with", then type "/users/". This generates a search of all pages of type https://edgeryders.eu/users/XXXXXXXXX, and you can see how many they are.
I think the way to go is to create system pages with different urls (for example “/start” instead of “/home”. Design them according to some hypothesis. For example: the hypothesis “people want to see activity streams more than they want to see featured content” would generate a home page based on feeds and streams of things that happen. This phase requires a lot of panelizing, view making etc. Once it is ready, we go through the procedure and serve the candidate home page to 50% of the visitors (treatment group) and the existing home page to the other 50% (control group). Then we track how the different homepages influence our goals, for example signup. GA does the statistics, and tells us when we the number of observations is high enough that it can declare a statistically significant winner.
So, any hypothesis?