After a 1.5 hour meeting with interested parties at SMX Seattle on June 3rd, followed by a second 1 hour meeting on June 4th, we have set the following research directions and priorities for the Dynamic Page Adaptation module:

...

• Allow users to create test campaigns, each campaign including the following configurations:

  • Selecting one specific "blog" (site area or site page) for the campaign to run on
  • Selecting one specific goal to measure success
  • Assigning alternative page layouts (through the use of evoSkins) to the campaign.

• Allow users to activate/deactivate each campaign individually at will

  • Later allow specific schedules

• Simplified tentative algorythm to use when a campaign is active:

  • Serve different layouts randomly for new users

  • Serve the same layout to the same user all the time

    • Use cookies
    • Use IPs
  • Record goal success based on current layout
  • Determine the best performing layout
  • Determine the error margins
  • Determine when one layout is clearly underperforming compared to the others and discard it automatically
  • Repeat until a winner is declared automatically

• Introduce multivariate testing

  • This will probably required to introduce conditional blocks within the use of a single skin
  • This requires additional research on multivariate testing

• Introduce testing on variable content

  • Posts may be marked up with alternative texts
  • Widget content may also be marked up with alternatives

  • In some conditions it may be interesting for the system to automatically spin content using synonyms in order to determine best converting copy.

    • This will probably require high traffic sites.

We will research these different options starting this year. However, as these areas are very experimental we cannot anticipate at which time we will have a final working product. If possible, some of these results will be included in the upcoming b2evolution v4.