Web analytics is a good tool to add to your user research toolkit. I occasionally teach workshops on the topic of using web analytics data for UX, and Iâ€™ve found it enormously challenging to pull together a half dayâ€™s worth of material that is understandable and useful to people. For the last two workshops Iâ€™ve done, there were three exercises for workshops participants to do together. I asked them to get into groups, look at the example data, and discuss the data. I included a few starter questions. In this post, I take a look at one of the exercises and discuss how I would interpret the data.
This exercise focused on data from the All Pages report in Google Analytics, which you can see in the screenshot, and which Iâ€™ve put in a table to make it a bit easier to read if you donâ€™t enjoy squinting at screenshots.
|Page||Pageviews (table is sorted according to this metric)||Unique Pageviews||Average Time on Page||Entrances||Bounce Rate||% Exit|
Questions and Definitions
I asked participants to look at the data and think about the following questions. Theyâ€™re not necessarily the sort of thing youâ€™d look at every time you analyze analytics data, but this is obviously a bit of an abstract situation.
- Which pages do users tend to view multiple times? Which pages do users tend to view only once? Why might this be?
- What are the pages where users spend the most time? What are the pages where they spend the least time? Why do you think?
- What are common pages for users to enter the site? Why?
- Compare the bounce rates of the most common pages for people to enter on. Why might some bounce rates be higher than others?
This isnâ€™t at all an exhaustive list of what youâ€™d analyze in real world situations, or even necessarily the sort of questions you would ask in every situation. Except in situations where youâ€™re just browsing the data to look for something interesting, as in this case, the specific data you analyze and the questions you ask will be driven by what youâ€™re trying to find out.
Before we move on to some interpretation of the data, we should probably look at some definitions.
- Entrances:â€‹ In the context of a single page, it means the number of people that entered the site on that specific page.
- Exits:â€‹ The number of people that left the website after viewing a pageâ€”that is, the number of people for whom this was the last page they viewed during their session.
- % Exits: Exits divided by pageviewsâ€”that is, the portion of pageviews where that was the last page that a user viewed during their session.
- Sessions: â€‹A session refers to an instance of a user coming to your website. If the same user comes to your website three times in one day, that counts as three separate sessions.
- Pageviews: â€‹The number of times people have gone to a page. If a user views the same page multiple times during their session, each time counts as a separate pageview.
- Unique Pageviews: Whereas a pageview counts every time a user goes to a page, regardless of how many times they do it, a unique pageview counts only the first time a user views a page. This metric gives you a sense of how many individual people have viewed a page, regardless of how many times they did it.
- Bounce:â€‹ When a user enters a site on a specific page, and then leaves the site without going to another page or interacting with anything on the page that you are tracking. In other words, itâ€™s when a user makes an entrance on a page and exits from that same page, without Google Analytics recording any other data about them using the site.
- Bounceâ€‹rate:â€‹The portion of pageviews that are bounces. Itâ€™s important to remember that bounce rate is only calculated based on entrances, and not based on all the people that view a page.
Unfortunately, thatâ€™s a lot to take in all at once, and analytics deals with a lot of concepts that human brains were never meant to understand. Thatâ€™s part of the difficulty of teaching a workshop on this topic.
In this section, I write up a high level analysis of the numbers in this report. I donâ€™t get too deep in this section, mostly because I donâ€™t have a particular questions Iâ€™m trying to answer. The biggest takeaway from this section should be that analytics data is a great way of generating research questions for you to answer.
Pageviews and unique pageviews
When you compare pageviews to unique pageviews, something that stands out is that no pages have outrageous ratios, implying that none of the pages on this site are ones that people reallyÂ view lots of times during their sessions. On other websites, you may find some pages that have a 2:1 ratio or more.
The home page has a ratio of about 13k:10k, which is on the higher end. This probably makes sense for a home page, since lots of people are going to treat it as a navigational hub. The â€œprogrammeâ€ and â€œspeakersâ€ pages, given that theyâ€™re a list of links to specific talks and specific speakers, respectively.
In comparison, pages like â€œvenueâ€ donâ€™t seem to have lots of repeated viewings, implying that users get what they want out of the page and then move on to other pages.
Time on page
â€œRegistrationâ€ and â€œinvoiceâ€ seem to have the longest times on page, consistent with there being stuff that users have to do on those pages (filling out forms, most likely). The average time on page for â€œvideosâ€ feels a bit short and itâ€™s worth exploring whatâ€™s going on with that page. Are there embedded videos? Can we look at stats on whatever service hosts the videos? Have we tagged the page to track outbound clicks?
â€œRequestâ€ and â€œcheckoutâ€ also have very long times on page. What are those pages about?
The average time on page for the home page actually feels rather high. It may be worth exploring through in-person research or maybe through a heat map tool to see what things people are actually spending so much time doing on that page.
Unsurprisingly, people enter on the home page. You may see different patterns on different kinds of websites, like ecommerce, where people predominantly enter on pages other than home.
The low number of entrances on â€œinvoice,â€ â€œrequest,â€ and â€œcheckoutâ€ indicates that they are probably pages from the middle of the transaction.
Pages about specific speakers also seem to have low entrances; my theory would be that this site just doesnâ€™t rank very high in search results for people looking for information on those speakers.
Three pages have bounce rates that feel quite high: â€œpractical-tourcity,â€ one of the speaker pages, and the videos page. However, the first two have very few entrances, meaning that bounce rate is calculated off of a small number of people. Iâ€™d be skeptical about whether bounce rate means a lot for those pages. For videos, though, it would be worth exploring the page and how people interact with the page to see whatâ€™s up.
The most interesting pages to look at for bounce rate are the ones with the most entrances: â€œhome,â€ â€œregistration,â€ and â€œprogramme.â€ The bounce rate for these three pages doesnâ€™t seem weird, really. It may be worth looking at how people interact with the registration page to see why it has the highest bounce rate. It could be that there is pricing information on the page and people are just trying to find that information.
Next Steps: After the Analysis
Again, this wasnâ€™t a rigorous analysis in large part because there was no real question I was trying to answer. So, of course, oneâ€™s next steps in real life would depend heavily on what youâ€™re trying to find out about users.
One idea would be to take a close look at the actual pages to get ideas about why people may be interacting with them the way they are. In some cases, you may want to do a path analysis to see where people are going from particular pages, particularly for pages where it looks like some people may be viewing them multiple times.
The biggest mystery pages are probably â€œregistrationâ€ and â€œvideos.â€ Iâ€™d check whether there is stuff on the page that takes people away from itâ€”maybe the videos are hosted on another site, or the actual registration takes place on a third party site.
At a high level, the three kinds of next steps that spring to mind are:
- More analytics data analysis
- Looking at the actual design of the website to come up with good stories about the data
- Research activities that donâ€™t include analytics to fill in some qualitative answers
Whether or not youâ€™d even do more research at this point would be a matter of what youâ€™re trying to find out and how important it is to reduce uncertainty. There are situations where the fast answers from web analytics is good enough; in others, you want to be as sure as possible about why users are doing what theyâ€™re doing.
Hopefully this has been a modestly helpful look at the way you might approach analytics data. Technically, itâ€™s not hard to start learning about web analytics tools, but learning about the data can be an enormous challenge.