Written by Robert Petković, Analytics Lead, Bruketa&Žinić&Grey Agency
If you are using Google Tag Manager to send your website data to Google Analytics, for the last couple of months you must have been looking forward to the new features added to Google Tag Manager. If you are not using Google Tag Manager, it is about time you started, and if you have no clue what Google Analytics is, then I believe you have come across this post by chance.
Google Tag Manager has recently introduced a few new types of triggers, which can help us track the activities of website users, which required our use of special instruction sheets before. The interface incorporates the following types of triggers:
- Element Visibility
- Scroll Depth
- YouTube Video
Figure 1: A list of triggers for scroll depth tracking and the position of new triggers in the Tag Manager interface
How can we use these new triggers? I will give some practical examples for each one of them, hoping they will prompt you to start using them.
This trigger is fired when some elements we want to track, either by tracking their CSS class or a specific ID of a particular HTML element, appear visible to a user in the visible part of the screen (Viewport). For instance, the moment when “footer” appears on the screen in front of the user, we can trigger an event that will tell us “The user has reached the end of the text.” We can also record when the user sees a banner (when say more than 50% of a banner was displayed to the user), and not only when a banner is delivered to the server.
Figure 2: An example of a trigger fired every time when more than 10% of the HTML element “footer” becomes visible on the screen:
By using such events, we can get more accurate answers to the following questions:
- Did someone read our text to the end?
- How much time did they need to read it?
- Did the user see the ad or the visual we delivered?
- How many ads do individual users see when they read our text?
- Do the users leave the page before we show them the newsletter signup form or after that?
Not only can we get answers to such questions, but based on such information we can also make the following conclusions, for instance:
- Does a banner displayed have any impact later on the sale of a product?
- Do the users read the most viewed text to the end or they only read the beginning?
- In the last paragraph of the native text, we singled out some brand benefits. Did the readers read that part at all and did that text affect their later interaction with the brand? Did they see the cool widget we implemented?
Such conclusions can help us better plan content modifications on our websites depending on what produces better results for our brand and what our readers better respond to.
This trigger is fired when the users start scrolling the web page. In this way we can see whether the users had any interaction with our web page i.e. whether they started to read the text and how much of it they managed to read at all. The trigger reacts to the percentage of page scrolling points, to pixel depth (e.g. “it is triggered when somebody scrolls more than 500 px”), and can also react to horizontal scrolling.
By implementing such trigger, we are able to record or find out the following:
- What percentage of the text is read on the average?
- Do the visitors from Facebook read only the beginning of the texts or they read more than a half?
- Who reads more content, the visitors from Google browser or from our paid campaigns?
- Our campaign has attracted a large number of users who lingered on the page for a while. Did they, while so doing, read the text to the end?
- Who achieves our web targets better, the users who read the text to the end or those who read only the first half?
- How much time does a reader need to read a half of this text?
- Do our texts need to be so long/short?
Such information can help us formulate the Content Strategy as in all foregoing cases we obtain specific information generated by real visitors, and not by us. We must not forget that we (as an agency or an advertiser) are too much involved in content creation to be objective enough.
As you may well suppose, this trigger is fired when somebody starts watching a YouTube video embedded into our website. Moreover, it can not only record whether somebody started a video but it can also capture the following:
- Video Start
- Video Complete
- Pause, Seeking and Buffering
- Progress or percentage/time threshold of video viewing
The following figure shows an example of how a trigger is set up:
Figure 3: GTM trigger configuration for YouTube
YouTube Video tracking also offers the following variables that can be used to arrive at better conclusions:
- Video Provider
- Video Status
- Video URL
- Video Title
- Video Duration
- Video Current Time
- Video Percent
- Video Visible
By using the above-mentioned trigger, you can for instance create an event entitled “YouTube”, and the Action could be something like “Video Start”, “Video End”, “Pause”, “25%”, “50%”, “30 sec” etc., and the Label could be the specific title of the video (from the “Video Title” variable) or the specific URL of the video (the “Video URL” variable). Both variables receive their values from the YouTube service, and thus the “Video Title” variable contains the title of the video featuring on YouTube.
In this way, you can get the answers to the following questions:
- Which video is most viewed on our pages?
- Which video is mostly viewed to the end?
- After how many seconds do the users stop watching a video?
- What is the total number of video minutes the users had in the previous period?
- Do the users pause our video with a cake recipe and at which spot? Is it convenient to us that they pause the video after so many seconds or at some other spot?
- Did the users who viewed the video make any conversion on our website? Which video has more impact on conversions and how? What should our next video be like in order to improve our website conversion rates?
If you have a website with embedded YouTube video content, such information can help you better design the forthcoming video content to be added to the website, to see whether it makes sense to have video content embedded into the website at all, to remove inappropriate video content from the website, etc.
The impact of events on Bounce Rate and how to use it
Given the fact that events have impact on the Bounce Rate metrics, my recommendation is that you should label such events at once as “Non-Interactive”. If you set them up as “Interactive”, the entire website Bounce Rate can easily drop to e.g. 5% and you start toasting because “you have finally managed to design such content strategy to have everyone read the texts”. Seriously, I have heard similar claims several times, and I know from my own experience that Bounce Rate under 20% is not really realistic.
We will all agree that there is however a difference in the level of interaction between the user who visited our website and “ran away” at once, and the one who visited our website, read the text to the end and left shortly after that. Such situations happen to us a lot, especially with users who visit us via social networks (that is why the “Social” channel often has a high Bounce Rate), and Google Analytics in both such cases suggests that there was a Bounce and that Bounce Rate was 100%. I think that it would be better that if somebody reads the text to the end or say more than a half of it, it is not considered a Bounce because the user has willingly had interaction with our text after all.
How can we then use the foregoing triggers to ensure a more accurate Bounce Rate?
The above-mentioned events can be turned into “interactive” so that they have a rightful impact on Bounce Rate in one of the following ways:
- “Interactive” events are triggered only when somebody scrolled more than 25% or 50% of the page. For instance, the event is not triggered when somebody scrolls 25% but only when they scroll 50%, 60%, 70%, etc. It is possible to trigger a special “non-interactive” event in only such cases when somebody scrolls up to 25%.
- “Interactive” events are triggered after only 5 or 10 seconds. In this case for a “Non-Interaction Hit” the variable is set up with the initial value of “True”, which becomes “False” only 5 or 10 seconds after the page has loaded. See Figure 4 for the example 😉
Why 5 or 10 seconds that is 25% of the page? Well, any activity performed by the user within 5 seconds or only at the top of the text page can in fact be regarded as a “chance” activity and it cannot be expected that the user clicked or scrolled something triggered by the content seen on the website. Anything more or longer than that can already be an indicator that the visitor had interaction with our content.
But that is not all (no, you are not going to get a set of knives now)! In addition to the above-mentioned, for Event Value I usually put the number of seconds that elapsed after the page has loaded in order to find out if on the average the users reached the middle of the text after 20 seconds or say 60. This information helps me analyse the progress of reading the text on the page so that I can give content authors more accurate information about the time needed by users to read such text.
Figure 4: An example of setting up an event triggering tag when somebody starts to scroll a page:
If you need technical assistance for setting such events up, do not hesitate to contact me.
The objective of digital (performance) marketing is not only to create a need for a product or to gauge how many users have seen the product, but also to detect how our digital activities later affect the sales or the use of a product or a service. The role of a good analyst is to use such information to offer creative people adequate guidelines for the next campaign or version of the creative solution. Without properly measuring the website activities of the users and reaching proper conclusions, we can only assume what impact our textual, graphic or multimedia content has on the user’s opinion about a product or a service, and we all well know that “assumption is the mother of all…” 🙂
On the other hand, the figures can show us a bitter truth about how our too awesome text or graphics are perhaps not so awesome to the users of a particular brand. A wise creative person who wants to learn and develop will take such information into account when creating the next campaign, whereas “others” will follow their own counsel until purely by accident they discover what makes the users tick.
Anyway, if you know exactly what you are going to do in your next campaign, it means that you are not ready for changes and that somebody else is already better than you. Why? Because they are trying out new opportunities and learning new specific information that will help them be more successful than you in a matter of time.
If you want to be a wise creative person in your line of work, make friends with your analyst 😉
First published on Bruketa&Žinić&Grey