The history of Web analytics tools has left a legacy of metrics that we can obtain "out of the box" even if they are of no practical use, and I would argue that a prime candidate for this category is time spent on page, and its troublesome partner average time spent on page. It's available because it's easy to obtain from tag-fires (or server log files) - it's just the time taken between consecutive page loads. Is it useful? Not by itself, no.
For example, it can't be measured if the visitor exits from the page. If a user doesn't load another page on your site, then there are no further tag-fires, and you don't get a time on page. This means that you have a self-selecting group of people who stayed on your site for at least one more page. It entirely excludes visitors who immediately tell they have the wrong page and then leave. It also, sadly, excludes people who consume all the content and then leave. No net benefit there, then.
Worse still, visitors who immediately realise that they have the wrong page and hit the back button are included. So, is there any value to the metric at all? In most cases, I would argue not, although there can be if handled carefully. For example, there is some potential benefit in monitoring pages which require data entry, such as checkout pages or other forms. In these circumstances, faster is definitely better, and slower suggests unnecessarily complicated or lengthy. For most shopping pages, though, you will need a much clearer view of whether more time is better or worse. In an international journey, four hours on an airliner is very different from three hours in an airport.
I mentioned that time on page is not helpful by itself: it can be more informative in conjunction with other metrics such as exit rate, bounce rate or revenue participation. For example, if a page has a high exit rate and high time on page, then it suggests that only a few people are finding the content helpful and are prepared to work through the page to get what they want - and to move forwards. Remember that you can't draw any conclusions about the people who left - either they found everything they needed and then left, or they gave up quickly (or anything in between).
So, if you use and quote average time on page, then I suggest that you make sure you know what it's telling you and what's missing; that you quote it in conjunction with other relevant metrics, and you have decided in advance if longer = better or longer = worse.
