Length and Depth of Engagement

In a previous post, I noted having some difficulty finding any studies that correlated the length of time a person spends on a page to whether they complete a transaction on a site - my notion being that designing pages according to the "average" length of time a visitor spends was catering to the attention given by non-buyers. I still haven't seen the numbers, but gathered some statistics on a non-commercial site to check the correlation between the length of time spent on a page and the depth of the engagement. Here's what I found:

The scattergraph above plots the number of seconds spent on the home page against the number of pages that a visitor viewed during their session. And while the results are fairly scattered, there does appear to be a significant correlation between the two: users who spend more time on a page tend to view more pages on a site.

I'm not entirely satisfied, though:

First, it's an analysis of a non-commercial site. My hypothesis is that, on a commercial site, the more time a user spends on a site, the more likely they are to complete a purchase during the same visit - so the behavior of users on a noncommercial site is necessarily a reliable indicator or behavior on a commercial one.

Second, I'm leery of implying a causal connection - that by "making" a user spend more time on a page, it will increase their engagement and make them more likely to consume additional content. I think the two are likely related, as this brief experiment shows, but any attempt to force a longer page-view is more likely to result in frustration than increased purchase completion. There's a correlation, but without an A:B test, I'd be very reluctant to assume a causal relationship.

Third, it's a fairly small data set - only about 500 visitors, during the course of a weekend, when they are logically more likely to be engaged in leisurely browsing than in active information-seeking. My sense is that this skews the numbers to more time and more pages in general. To be firmer in the conclusion, it would take a larger number of users over a greater length of time.

All in all, it's a very informal analysis - but it points in the direction I anticipated, and suggests that further research might be worthwhile.