As a result of these developments, several issues deserve further investigation. Firstly, it is clear that clickstream data can only partially reflect the effectiveness of a site. Clickstream data display the ‘when’ and ‘what’ of web visits, but are of limited use to answer ‘how’ and ‘why’ questions about customers’ site use. Secondly, despite their apparent richness, clickstream data have some serious flaws. Often, they do not cover all site traffic in complete detail, and managers need to be aware of the gaps in the available data. Thirdly, since academic researchers tended to focus on customer surveys, our understanding of how managers are actually using web metrics for site improvement is rather limited