16th November 2020
Bounce rate, time on page, new vs returning users, open rates, reach, impressions... We look at popular metrics that are commonly misunderstood, leading to inaccurate data analysis.
Digital might be in our DNA, but even we recognise that there is a lot to take in when you begin looking at analytics reports. The sheer amount of numbers and the somewhat questionable naming conventions between analytics reports can be overwhelming. In fact, trying to include the different metrics and platforms in one single piece of content is equally overwhelming! To help make the topic as digestible as possible, this series of three articles will look at examples across:
- Website Analytics
- Email metrics
- Social metrics
While there are many platforms to choose from, Google Analytics remains the sector leader thanks to the tremendous amount of data it can capture on your website including visitor behaviour, demographics and device information. What’s more, it seamlessly integrates with other Google marketing platforms such as Google Ads, Google Optimize and Google Search Console. To top it off, it is free, making it the ideal solution for many websites. There is a paid version, starting at a rather steep $150,000/year; Google Analytics 360 is designed for websites that demand a more robust version of the platform with advanced analytics and data-driven attribution.
Google Analytics 4
At the time of writing, Google announced the launch of Google Analytics 4, the latest iteration of Google’s web analytics platform. This newest version gives you predictive insights, deeper integration with Google Ads, cross-device measurement capabilities and more granular data controls.
With so many businesses using Google Analytics, this is exciting news as the new property provides a more comprehensive, cross-channel view of the customer lifecycle and uses this information with its predictive marketing features to provide marketers with more insights.
Google Analytics 4 will be the default option when you set up a new property. However, the previous iteration, Universal Analytics, will continue to remain available and Google recommend that site owners set up both property types and run them in parallel. It is also worth noting that new feature developments will be focused on Google Analytics 4 and we may well be in a position where Universal Analytics is made redundant by Google, enforcing a switch to GA4.
Alternatives to Google Analytics
Although it may be one of the most used website analytics platforms, several businesses choose not to use Google Analytics due to concerns of data ownership and use of those domains fortunate enough to hit the sampling thresholds. You can find more information on Google Analytics sampling thresholds here. There are a great deal of other platforms out there, some of our favourites include Adobe Analytics, Clicky, and Kissmetrics.
Using more than one platform
While there are plenty of commonalities between these platforms, there are some key differences that any business needs to consider before choosing which platform best suits their needs. Some websites choose to implement a combination of these platforms; usually, the free version of Google Analytics followed by one of these premium solutions. This is perfectly fine, however, you need to be aware of the differences in how they collect and process data to explain the probable differences in numbers.
Let’s have a quick look at a key difference between Google Analytics and Adobe Analytics.
Google Analytics Sessions vs Adobe Analytics Visits
The first notable difference is the decision to name “visits” to a website as “sessions” within Google Analytics. Visits, or sessions, are a group of page views made by the same user in a short amount of time. Visits on both platforms typically expire after 30 minutes of inactivity. Several scenarios can cause differences on each platform.
- End of day: All sessions in Google Analytics expire after 11:59 PM. If the user is still active on your site after 12 AM, a new session is created; Adobe Analytics count visits that continue into the following day as part of the same visit.
- Different campaigns: In google Analytics, a new session begins if a user's campaign source changes; for Adobe Analytics, it is considered to be one visit, even if a new Tracking Code value is seen.
Getting to know the metrics
If we were to build a glossary of metrics, we would have over 100 metrics each for Google Analytics, and Adobe Analytics - just 2 of the several web analytics platforms on offer. For now, we’ll focus on some of the most misunderstood metrics.
Sessions/Visits - a single visit to your website, consisting of one or more page views. The default visit expiry is 30 minutes, which means that if someone is inactive on your website for over 30 minutes, a new visit will be reported if they perform another interaction, for example, viewing another page.
Visitor/User – another difference in naming convention between platforms, this metrics shows a person browsing your website.
Bounce & Bounce Rate – a bounce is when a visit consists of a single pageview. Bounce rate is the percentage of visits with a single pageview.
Time on Page/Site – shows how long visitors are spending on your website and individual web pages.
Event - a custom interaction that is tracked from your website, for example tracking form submissions, or tracking plays of an embedded video. These are custom attributes that need to be tailored and configured to track, as they are not set out of the box.
Now we have a base level of understanding of the key metrics, let’s explore some common misunderstandings when reporting on these.
High bounce rate is bad
Hold on a second, before we make that assertion, let’s ensure we have fully understood what a bounce is, and put it into context. We mentioned above that a bounce is when a visit consists of a single pageview. What we haven’t touched on yet are the default metrics recorded by web analytics vs what you need to configure using events, that can have a significant impact on the number of bounces recorded.
By default, web analytics cannot track on-page interactions. Therefore, a bounce means a user has only viewed one page and left in the eyes of the web analytics. For many people, they think this also means the user was dissatisfied with the page which is why they have left. This may not always be the case. For example, say you have a single page website, there is no other page to navigate to – which means that every visit is a bounce. These style pages typically have top-level navigation that scrolls you down to a specific section on the page; an interaction that will not be recorded by default. Go a step further: say your landing page has a video embed; the number of plays or how long the video is watched are additional metrics that are not tracked by default. Without collecting these interactions, a potentially high performing page can result in a high bounce rate being recorded and can lead you to believe the page is underperforming.
Another issue with the default bounce metric is that time on a bounce page is always set to zero. Time on a page/site is calculated between page views- more on this a little later. As a bounce is technically a single page view, the time on page is set to zero. Ok, but what if a user has spent a considerable amount of time on the page reading the content or watching an informative video; why should this be set to zero and recorded as a negative signal. People can read your entire page, spending 5, 10 or 15 minutes reading every word, but if they don’t click to another page on your site, web analytics considers it to be a bounce.
As you can see from these examples, the bounce metric is becoming a low-value metric as it is no longer providing a reliable picture of the engagement or satisfaction people had with the page. It only shows that the page did not drive users to view another page on the website within this visit, ignoring the quality of their interaction with that single page.
How do we fix this issue?
To get the best idea of your website’s performance, you need a solid understanding of what high engagement and user satisfaction is for your website, in fact for your pages. Segment your site into page types such as: homepage, campaign landing page, article page, contact us page etc. Once you have this, map out the metrics you need to collect to accurately assess the performance of the page. Take an article page such as this one you are reading now. The purpose of this content is to educate and encourage readership. We will therefore be tracking a number of custom events to help us understand how engaging the content has been:
- Scroll depth
- Time on page
- Link clicks within body content
- Navigation clicks
These will be used in conjunction with the default metrics recorded by our web analytics.
So, is bounce rate a poor performance indicator?
If used on its own, without defining what engagement and user satisfaction means for a given website/web page, then yes, bounce rate is a poor indicator. That said, if used in conjunction with other metrics such as the events mentioned above, you will begin to form a real picture of how a page is performing.
0 time on page = no reads
As mentioned earlier, time on a page/site is calculated between page views. However, seeing the time spent on your website or on individual pages is not as obvious as it first appears; you may have already experienced the head-scratching scenario where your average time on site is lower than the average time on page.
Web analytics by default cannot measure the time a user has spent looking at the last page of their visit. This happens because web analytics platforms use the time of the next page view to determine the time you spent looking at the current page. On the last page, there is no next page recorded, so the time on page is unknown, thus recorded as zero, and the visit duration ends when at the point that the visitor opened the last page. This table should explain it a little clearer:
|Page viewed||@time||Time on page||Visit duration|
|Homepage||09:00||09:30 – 09:00 = 00:30|
|Article page||09:00||09:30 – 09:00 = 00:30|
|Homepage||09:30||09:55 – 09:30 = 00:25|
Let’s use Visitor 2 in the above table for our first scenario. You have created a carefully crafted article page full of useful information on topics identified as a hot subject for your audience. You spend hours putting this together, content + design + implementation and market the content externally. You attract Visitor 2, who spends a considerable amount of time reading the content, getting answers to a subject they wish to learn more about, and then leave the site fully satisfied with their experience. The default web analytics tracking will record this visit as a bounce, resulting in the time on page and visit duration being recorded as 0. This time reading is clearly because the analytics could not calculate the time and not an accurate reflection upon the user journey.
Luckily there are several techniques to use in order to see if people are actually reading your content, even if they bounce.
How do we fix this issue?
This conundrum can be fixed by introducing events, a custom interaction that is tracked from your website. We have already mentioned how tracking events can impact the way in which bounces are recorded. The same applies to calculating time on page/site. By tracking events, you can change the way visit durations are calculated by enabling analytics to look for the timestamp on the last interaction hit. This means if you track things like video views, and the visitor starts viewing the video on the page, the visit duration is calculated to the time of that event.
How can you use time metrics?
The same applies to time on page/site as for bounce rate: use these metrics in conjunction with other metrics to form a real picture of how a page is performing. We tend to recommend looking at time metrics with bounce, exits and events together for a given page type.
A Visitor/User is a single person?
Wrong. Technically speaking, this metric actually shows a unique browser cookie browsing your website. For example, let’s say our friend Mr E is browsing the Volume Up site on both his phone and laptop, the web analytics will record these as 2 visitors/users as there is nothing signifying that this is the same person.
Now the real issue with this metric that so many get wrong; you cannot simply add up visitor numbers and expect the total to match a total visitor count for your website. Let’s use the below simple scenario to explain in more detail:
As you can see this website has been recording 100 visitors every day for 10 days. When totalled the sum is 1000, however, the report shows the total visitor count for the 10 days is 920. That is because 80 people visited on numerous days within the date range being analysed. You should never add visitor numbers together to get a total. Instead, establish the date range you are trying to analyse and let the analytics platform display the correct visitor count. We tend to recommend including monthly and year to date visitor counts for reports to give an indication on how you are performing for those given periods. In most cases, you want to build loyalty and entice return visits. Any retention-based activity will therefore have an impact on your overall visit counts as you begin to build a loyal fanbase who are returning and engaging with your brand on an on-going basis- so high numbers of new visitors aren’t necessarily a good thing if you don’t also have high numbers of revisits.
Without a solid understanding of what good looks like for your page/site or marketing activity you are reviewing, you will not be able to make the right decisions on how you are performing or where you should be allocating time and effort to improve performance.
With an impressive depth and breadth of expertise across industries, our team knows what to measure, and how, to help clients to get maximum impact from their data. Our Smart Health Insights (SHI) for Healthcare programme enables clients to quickly and easily see a tailored view of their performance and offers guidance on the best way to improve it. One of several ways we give our clients a head start in their journey towards truly understanding data.