Measuring the effectiveness of a portal initiative has always been a long sought-after goal. The real to measuring effectiveness is to get people to agree on what it means to be effective. Is that number of page hits? length of time on the page? increases in order activity?
Pradeep Behl at IBM published a white paper on DeveloperWorks that attempts to explain how to use web site analytics to measure Portal effectiveness. The paper is titled “Measuring IBM WebSphere Portal effectiveness” and was published on 7/13/2010.
In the article, Pradeep monitors a predetermined set of activities measured the actual impact (page hits) against an anticipated impact determined prior to roll out. This activity impact ratio (actual/anticipated) is then adjusted by the relevance of each activity to the overall program.
For example, a particular page on the portal may represent 10% of the activities of our new portal. We anticipate 10,000 page hits and actually get 2,000 hits. So our activity impact is 2,000/10,000 = 2. Multiply 2 * 10% and we get 20% effective rate.
We must then combine our activity impact with other measures, such as the amount of problems reported, scores from user satisfaction surveys, etc. Once we average out all these measures, you can come up with a total effectiveness score.
Below is a process flow from the article. 
The article proposes an interesting approach to measuring effectiveness of any initiative, including portal. But as I mentioned at the beginning, the real hurdle is getting people to agree on what should be the measures and what initial goals are realistic.
