Starting with behavioural analytics
The key concepts to understand website reports
Quantitative data and especially behavioural analytics can complement qualitative research to uncover precious insights about who your users are and how they behave on your website. In this article, I want to give an overview of what behavioural analytics is. I also suggest ways they can complement user research and the design process. Finally, I describe the most important terms and metrics to get started.
“Data will talk to you if you’re willing to listen”
J. Bergeson
What’s Behavioural Analytics and what’s its value?
Behavioural analytics collect and store information about the behaviour of users on a website (or app). You may find sometimes it under the terms of web analytics or simply analytics. It works by introducing inside each page of the website a piece of code that sends information to an analytics tool. This process is called tagging. At the basic level, the information collected includes at least what pages were loaded, when, and from where the user arrived. Cookies are also involved in the collection of data as they store extra information about the user and their behaviours.
While common usability testing methods answer “why” questions, behavioural data are well suited to answer “what” questions. For example, “What pages are most viewed?”, “From what part of the world comes the majority of users?”, or “What people search inside the website?” [1]. For this reason, analytics are not meant to be used alone, but rather to complement other user experience research methods [2].
There are many ways behavioural analytics can be helpful to the point that Hay suggested in his book an analytics-first approach to improve the user experience. One excellent example is given in a post by UserZoom, who shows a data-driven approach to the creation of personas (it will be hard to argue against such well-grounded personas!). Analytics can also help to identify potential issues in your website experience and suggest where changes are needed by telling how users engage with the site content. For instance, it is possible to discover that many users are dropping out in the middle of a task (e.g. the sign up form) and so the redesign of some parts of the website might be beneficial. At later stages of the product development, advanced analysis such as A/B testing can optimise the website experience. Importantly, analytics data are collected constantly, so they offer the chance to monitor changes in users’ behaviours over time. For this same reason, analytics can also be used to measure the success of design decisions, by evaluating how users interact with the website following a before-after approach.
Nowadays there are several tools available to track and monitor the behaviour of users on websites. Two common solutions are Google Analytics and Adobe Analytics. Less known but equally valuable solutions to check out are HotJar, Kissmetrics, Matomo, and Heap.
When I first started reading about analytics I found myself amused by its potential to uncover behavioural insights. However, I noticed online learning resources often adopt a technical language which might steepen the learning curve. I compiled a list of the most important terms along with their explanation in plain language. I avoided referring to specific tools as much as possible, meaning the concepts you will find will be equally valid in any analytics product you might choose to employ. I also added the key metrics to master to get started.
Key terms and metrics to master
Metrics are measurements, or quantitative data, collected when users browse a website (or app), like the time spent on a webpage, or the number of downloaded documents.
Behavioural analytics tools offer tons of possible metrics that can help in understanding the user experience and business performance. The right metrics to analyse and monitor varies accordingly to the website type and business goals. For example, while it makes sense to measure the average order value on an e-commerce website, the same metric certainly makes no sense when evaluating the experience of a blog. Usually, the basic information to look at when starting includes the pages viewed by users, when the pages were visited, and the order of pages visited.
Always remember that the interpretation of each of these metrics will depend on the characteristics of the website. Numbers by themselves are neither good nor bad: the interpretation of a metric is always left to the researcher. For instance, a high time on website pages is a good thing for a blog because it could mean people enjoy the contents, but the same is a bad thing for the results page of an e-commerce platform because it could mean people are having troubles finding what they are searching. Note that I am saying “it could mean”: when dealing with analytics you can only make assumptions about the reasons behind users’ behaviour. To uncover them, one must resort to more appropriate research methods, like interviews.
1. Sessions (or, visit). A session measures the time a user spent browsing the website. When a user has not visited any page for more than a certain amount of time the session stops counting (usually the default is 30 minutes of inactivity). If that same user enters the website after this time limit, it will be recorded as a new session. For instance, if someone visits a website for 20 minutes, then leaves for a 1-hour lunch break and later enters once more into the website, that will be counted as 2 sessions.
2. User (or, visitor). The person browsing the website. However, always bear in mind that a “person” corresponds to a unique cookie. That means that if a user clears the browsers’ cache and then enter the website, it will be recorded as a new user. This also means that the same person accessing from different devices may be counted more than once (because each device has its own cookie).
3. Unique users (or, unique visitors). The number of users who created at least one session during the day. Remember that a user is only unique within the time framework selected (usually, 1 day). If the timeframe is set to 1 day and the same user accesses the website every day for two weeks, it will be counted for each day as a unique user; if the timeframe is 1 week, then it will be counted as a unique user twice (1 for each week). Again, keep in mind that users can access the website from different devices and so they will be counted from each device as unique users. Knowing who comes back regularly can provide useful information about what type of users the website is attracting.
4. Pageviews. The total number a page was viewed, including repeated visits to the same page. For example, if the same user visits the contacts page 3 times than pageviews will be equal to 3. This metric is useful to understand what pages are driving users’ interests. It can also be used to identify problematic pages, meaning pages that received only a few visits: you may want to ask why is so and start investigating by planning some interviews.
5. Unique Pageviews. Similar to pageviews, but it excludes repeated visits. So, if a user visits the same page more than once it will still be counted as 1 unique pageviews for that page.
6. Hit. It refers to any interaction that generates data. It’s good to know this term to avoid possible misunderstandings in the future. Technically, a hit is a request to a web server for any object to present on the website. Loading a page, sharing an image, downloading a whitepaper are all examples of hits. In advanced analytics, you may want to track specific hits, like social interactions or request for documents.
7. Pages/session (or, page depth). The average number of pages visited divided by the number of sessions. Pay attention to interpreting this value: a high value could mean people are interested in exploring the website; however, this could also be a sign that users are not finding what they are looking for and so the map site should be redesigned.
8. Average session duration. This metric tells how long on average a session lasts. Note that the time spent on the last page is not counted, so if a user enters and then leave the session duration will be equal to zero. Remember also that a user can create more than one sessions over a certain timeframe. It is useful to know that the average session duration is something less than 2 minutes and a half according to a 2016 survey conducted by Brafton.
9. Bounce rate. Bounce refers to a user who enters the website and then leaves without visiting any further. So, bounce rate describes the ratio between users who entered the website and then left divided by the total number of access to the website (i.e. the number of sessions)[3]. It is usually expressed as a percentage. The bounce rate is typically assessed against the landing page. A low bounce rate is, generally, a mark of successful design. Following Beasley (2013), a bounce rate lower than 30% is considered quite low.
10. Percentage of new users (or, visitors). This tells how many new users visited the page within a certain timeframe. It is computed as the number of users who never accessed the website divided by the total number of users in that same timeframe. Again, remember that users who access from different devices will be recorded as new visitors unless there are other methods of tracking (e.g. accounts). Examining the arrival and characteristics of new users is especially interesting to look at if a marketing campaign was recently promoted.
11. Users Flow. Some analytics tools are able to visually represent how users move across the pages. The user flow shows the paths users follow when browsing the website. Notably, it provides information about the percentage of users leaving the website from each page.
12. Active users. The number of users navigating the website in real-time. Note that the meaning of real-time may vary with the tool used. For instance, Google Analytics include data for the previous 30 minutes.
13. Active pages. The pages of the website users are navigating in real-time.
14. Exit rate. This metric shows for each page the percentage of users who left the website after reaching that same page. Similar to the bounce rate, but different: while the bounce rate account for the number of people who entered the website and then left, the exit rate includes users who browsed at least some sections of the website and then left. Looking at the exit rate may help to pinpoint what pages are problematic.
15. Conversion rate. Conversion is a generic term for all those user activities that contribute to the success of a business. So, there is not a unique definition, it will depend on the business and its objectives. For example, it could refer to signing up for a newsletter, sharing a story, or completing an order. For this same reason, the way to compute the conversation rate may differ, but it is generally calculated as the number of conversions that occurred divided by the number of total possible interactions tracked in the same time period. It is usually expressed as a percentage. When the conversion rate is analysed together with other information it can generate powerful insights. For instance, let’s say the conversation rate for signing up for a newsletter is 15%. It is possible to differentiate the signup rate depending on the device. One might find that users who signed up using a laptop had a 10% conversion rate, while those who signed up using a mobile device had a 20% conversion rate. This gives a very good idea of where to focus the efforts. If you are wondering what would be a good conversion rate, CrazyEgg’s blog wrote an extensive article with some free advice.
Final thoughts
Behavioural analytics is a powerful source of information about your users and their behaviour. However, when starting the technical language can create some frustration and discourage learning. In this post, I hope I shed some light and clarified the most important ideas without being tied to a specific tool.
The terms and metrics described here barely scratched the surface of the power of behavioural analytics. Especially in relation to e-commerce, such data can become extremely helpful even to the point of revolutionising the business strategy. Recalling Bergeson’s quote, if you are willing to put your efforts and start understanding data, data will tell you precious things about your users. Nonetheless, let me stress one final time that analytics do not substitute but complement other UX research methods.
Footnotes
[1] What people search inside a website is actually part of a sub-topic of analytics called search analytics (not to be confounded with traditional SEO analytics).
[2] See especially the triangulation approach described by NN/g.
[3] Google Analytics Help provides another way to think about the bounce rate based on sessions.
