I often post to the official Google Analytics User Forum – this post is an expanded response to a question I answered on the forum.
"I've recently launched a big redesign on an ecommerce site. After that, the Avgerage Time on Site increased by around 1.5 minutes, and the Pages/Visit dropped by half. Now I could explain the Pages/Visit to making the site's structure easier and faster, now the customer finds the product he wants with less clicks. But then why is the Average Time on Site increased? The combination of these two statistics is what makes me think that there's a logical gap here.
This is certainly an interesting change in your data. There are two possible causes:
- Something broke in your analytics implementation (or unbroke) with the launch of the new site.
- User behavior has changed due to the redesign.
To understand which one is the case here, let's take a step back and review how Google Analytics works.
Metrics, like time on site and pages/visit, are session based ratios. These numbers will change when the input for the calculation changes. Both depend on the detection of a session (visit) in your data. If you see session-based ratios change, it means either the numerator (pages viewed in the session, total time of the session) has changed, or the denominator (sessions aka visits) has changed relative to the numerator.
Both the numerator or denominator can change as a natural result of the redesign. If something was broken in your GA data before and fixed now, or something was broken with the redesign, the most likely change would be the denominator, though the numerator could change in one case: that being where your old site over-counted pageviews.
If something has broken, you're likely to see swings in other session-based ratios and sessions themselves: look at bounce rate – if it has spiked too, that's a problem; look at total visits – if they've jumped, that's also a sign of session corruption due to something about your GA install being broken.
Possible Answers to the Time on Site Change
- Your old site over-counted pageviews and your new site's GA implementation corrects this. If this is the case, it would be expected to see (a) Visits remain the same, (b) Time on Site remain relatively the same, and (c) Pageviews decrease and thus pages/visit decrease. Why would your old site have over-counted pageviews? Perhaps it used virtual pageviews for things like outbound links, downloads, or other events rather than Event Tracking. Perhaps some pages were double-tagged, resulting in two pageviews being reported on those pages when only one should have been reported.
- User behavior has actually changed. It's possible that pages/visit would go down with a redesign while users' total time spent on the site wouldn't change much and visits wouldn't change. This would, in my opinion, be a sign of positive things. Quite honestly, 17 pages/visit is NOT a good sign in my experience. For average sessions to have to view 17 pages is crazy high – most sites average is around three to four pages/visit. Higher pages/visit doesn't necessarily mean good things – if people have to view 10 pages to do something that should perhaps take two or three pages that's not good! Pages/Visit = good runs on more of a bell curve – too low and it's a sign people just hate your site and leave; too high and it's a sign people hate your site after trying really hard to find/do what they wanted.
So, what's the answer?
Check your data.
- If it's option one, there's something to fix.
- If it's option two, I'd say your redesign is probably a huge success!
What to do next?
Yours is a GREAT case for getting some better data – data that Google Analytics can't ever tell you: voice of customer data. I recommend looking into these two tools:
- 4Q Survey – a great tool for qualitative analysis – this may help answer the questions that GA can't about your data: why are people behaving in the way they're behaving?
- User Report – this is a free tool I recently discovered that's quite neat. It gathers demographical data and user feedback, all in one.
I hope this has been helpful! Feel free to post questions below.