Google Analytics shows you lots of useful data about your site’s usage and performance. However, not all of the metrics reported by Google Analytics are 100% accurate. At Whole Whale, we like to call these ‘Misleading Metrics’.
Keep reading to find out which metrics are misleading, why they’re misleading, and how you can use this knowledge to better inform your analysis.
1. Your Engagement Metrics may not be what they appear
Your average session duration is probably underestimated. Unlike a shop owner who can easily monitor when patrons enter and leave, GA relies on an algorithm to calculate the length of your session. GA measures your session duration by looking at the time-stamp of a page request. For example, if you pull up a site’s page at 7:05, and then request a second page of that site at 7:06, Google will subtract these two times and record your session duration as one minute.
At this point, you must be thinking, what if I only visit one page?! How can GA calculate my session duration if I never visit additional pages? You’re right. It can’t. Because there is no second time-stamp, a visitor who bounces (leaves after one page) is assigned a session duration of 0. 0 minutes and 0 seconds. It’s as if the visitor never came at all. This skews your average session duration lower than it should be. The formula for average session duration is:
Average Session Duration = Total Session Duration / Total Sessions
Bounced sessions add 0 in the numerator, but still count in the denominator. One way to get a better representation of session duration is to apply the ‘Non-bounce Sessions’ segment to your data. Look at the discrepancy between ‘All Sessions’ and ‘Non-bounce Sessions’ for this account. Filtering out bounces paints a much more accurate picture of what’s happening on site.
Ahh! Bounced visists are counted as 0 second visits on @GoogleAnalytics Click To Tweet
The next engagement metric is average time on page. This is a page level metric and a reliable measure of engagement – sometimes. We covered how GA calculates time on page: time-stamp of next page minus time-stamp of current page. For pages with low exit rate (% Exit), this metric is great. However, average time on page corrects for exits by removing them from the calculation:
Average Time on Page = Time on Page / (Pageviews – Exits)
If a page has a high exit rate, we lose all the data from the exits. Were those users spending 10 seconds on the page? 20 seconds? 20 minutes? The world will never know. Keep exit rate in mind when analyzing average time on page.
Pages per session is another default engagement metric offered by GA. This one is less open to interpretation. It’s defined simply as the number of pages loaded during a single session. No data is lost here, regardless of bounces or exits. Of course, the value of this metric depends on your site’s structure. If you offer content heavy pages that encourage more reading than clicking, pages per session won’t help you understand engagement.
Google’s out of the box engagement metrics leave something to be desired. To better understand how users interact with your site, set up custom event tracking and, most importantly, Google Analytics goals. Customized implementations have more reliable tracking and they’re specific to the experience you want your users to have.
2. Your organic search is probably overestimated
Aside from the rampant prevalence of keyword (not provided), another hassle with organic search keywords is the presence of branded keywords. After all, these people should really be classified as direct traffic! Think about it. Anyone who types your organization’s name into a search engine isn’t really using it to discover your site. They are simply using it as a means to navigate to your site, because they already knew they wanted to go there.
Fortunately, “organic search” data is customizable. Google Analytics allows you to modify search engine lists and more importantly, exclude keywords. You can exclude searches of your site as organic search, and these visits will be counted as direct traffic. Still no help with keyword (not provided) 🙁 Make sure to verify your site in Google Search Console and associate it with your GA account to get access to the Search Console data.
Remember to tag branded terms in your @googleanalytics as direct traffic! Click To Tweet
3. Count me in…twice?
Google Analytics keeps track of your site’s users by assigning cookies to your browser and device. The first time you visit the site, you get a cookie (not the sweet kind) which is active for two years. Then, with every subsequent visit, this cookie gets renewed and a new session is counted- but not a new user.
This method of tracking users can introduce some problems. For instance, since cookies are assigned to a unique browser + device combination, a single individual can be tagged with cookies on multiple browsers and devices. Following this principle, one user would be recognized as multiple unique users. Conversely, if multiple people share a device, it’s likely that they will simply be recognized as one user.
For logged in users, Google Analytics has a more user-centric approach to tracking. A user logged into Chrome, for example, is tracked across browsers and devices. If your site requires a login to access resources and content, you can make even better use of GA’s User ID feature. Overwrite Google’s default User ID values with your own, and set up a User ID view to see how these visitors perform differently from non-logged in users. When setting your own tracking IDs, remember to anonymize – Google has a strict policy about protecting PII (Personally Identifiable Information). User ID tracking helps to mitigate the effects of cookie overlap and deletion, giving you a more accurate representation of new and returning users.
4. Your direct traffic might not be, well, direct
Direct traffic ought to consist solely of people who type your website into their browser. We don’t want to count traffic that gets to your site via a referral link, search engine or newsletter as direct. Sometimes, it happens.
Bad campaign tracking code is probably the biggest culprit here. If you’re trying to track a campaign- be it email or referral- but your code has some bugs or can’t be read, Google will store the traffic as direct. Avoid this mishap through using our super helpful UTM builder spreadsheet.
URL tagging is also useful for capturing visitors who came from offline sources, such as a powerpoint or .PDF. Traffic from these sites will automatically get siphoned into direct traffic. You can campaign tag traffic sources to sort into more appropriate categories. If you haven’t noticed a trend here, it’s that campaign tagging can be extremely useful in illuminating the accurate source of your traffic.
It’s also important to have proper filters set up to ensure that internal traffic isn’t being counted. As OrbitMedia mentions, if employees have a company site as their homepage, direct traffic totals can be very misleading. One solution for this is to exclude the IP addresses of your office and team members. This will prevent internal traffic from tallying visits within analytics.
5. Bad Code Leads to Bad Data
Using sloppy tracking code can also cause problems. For instance, if you accidentally put multiple tracking codes on your page, your metrics are sure to be misleading. Similarly, improper placement of tracking code can also screw with your numbers, as well as impact page load time. Take the time to revise and recheck your code to make sure it’s executing properly. Don’t set it and forget it! Google’s Tag Assistant plugin helps you identify and troubleshoot your tags.
Bonus: Wait for it…
This isn’t really an inaccuracy in the data, just a caveat. Any data you see which was collected within the last 24 hours is unreliable. When setting your date range for analysis, be sure to exclude the current day. Google has tons of data and metrics to compile and organize, and if your site has heavy traffic, give Google atleast 8 hours to complete this process. Don’t worry about your Real Time Reports, though. That data can be counted on to be counted correctly. Pardon the awful wordplay.
Should you ditch Google Analytics? No way. The data discrepancies discussed above aren’t meant to undermine GA’s utility, but to provide some insight into how metrics are calculated. After all, we do not need perfect accuracy for Google Analytics to be useful. For instance, even if we know that the system for tracking users is flawed, we know that more users in September than in August is still a good thing.
And of course, each iteration of Google Analytics makes improvements so we can expect these problems to be minimized in the very near future.