Google Analytics Intermediate Review

I am doing a "Growth Marketing Mini-degree" scholarship at CXL Institute. And I will be writing a weekly review of what I am learning along the journey.

Here's my review for week 9

This week I will cover what I've learned in the "Google Analytics Intermediate" course.

The course went through:

? Cleaning data by filtering out spam, filtering internal hits

? It then scratched the surface of cross-domain tracking.

? Talked about funnel visualizations vs goal flow reports.

? Building segments from scratch, using sequences in them, building dashboards, building custom reports.

? Channels and different ways to make channel grouping.

? Multi-channel reports, how they work in a different way than other standard reports, and the value of using each report of them.

? Finally, the course came to an end with some tips and tricks on different subjects like attribution, content grouping, measurement protocols, annotations, etc.

Clean Data: Filtering Out SPAM:

? In GA, everything that is recorded, is not necessarily from an actual user, that's called spam.

? Dirty data hides the story that GA is trying to tell you.

? Spams are injected into your GA through "measurement protocol" which is how google analytics get his data.

? You can look at some types of spam data in "referral" under all traffic; websites that look like blackhat, seo.xyz, etc. Or in under audience -> GEO -> language report.

To get rid of SPAM data:

? Use "bot filtering" under view settings.

? Set a filter for "sources" that are famous for being spam.

? Use filter "include" for "custom dimensions", after setting them in GTM.

? Use a filter "include" hostname

? Use a second property.

Clean Data: Removing Internal Hits:

? Why should you remove internal hits? Because it would skew your data.

? GA doesn't show IP addresses in the reports but it knows them.

How to remove them (you can use any or some or even all of them):

? GA opt-out chrome extension.

? Make a filter exclude for your city if it's a little one with no possible traffic.

? Make a filter exclude for your IP address, but only if you are working in a large corporation where the IP address is known and fixed.

? Make a filter exclude source = test, but according to that filter, you should manually tag yourself each time you are testing something.

? Another option is to set a filter to exclude the custom dimension "original UTM source", but you should've been set it using GTM before that.

Clean Data: Cross-Domain Tracking:

? Without cross-domain tracking, every transition to a new domain is treated by GA as a new user with a new "client-id" is entering the site in a new session.

? Cross-domain tracking is involving the root domain and top-level domain, not sub-domains. So in sub-domains, GA and GTM will figure them out by themselves with no need for a pre-setup.

? One way to know you have set cross-domain tracking or someone is using it is the "ga=" followed by multiple numbers, in the URL, that's what is called a "linker" or "decorated link".

? That "decorated link" is passing through the "client id" to the other domain telling GA that is the same user in the same session.

? Another way to know if a website is using cross-domain tracking, is to use the "GA debugger" chrome extension and notice the client id that is used.

? To check if your cross-domain tracking is set right or if it causing you an issue, you can either use the debugger or look for self-referral in either your "source/medium" report or eCommerce "product performance" report.

How to set cross-domain tracking:

? You can use GTM, a simple one.

? Or you can do it manually on GA, but you will need to adjust some code on your pages.

? We have another issue similar to cross-domain tracking but not exactly the same; some serving websites like PayPal. Here you can't set cross-domain tracking because you even don't get referrals to PayPal at your GA account, it's not your website. So what will GA report is that a user with a specific Client-Id come from some source -say Facebook- then he closed the session -actually he went to the PayPal website-, then of a sudden, he appeared again after a while on a new session, referred from a website called PayPal and completed the purchase.

? What you are telling GA in "referral exclude" is: don't attribute any traffic to PayPal, just go back into that user's record and see what was the last traffic source was then give this source the credit for traffic.

? So if you have multiple root domains then you should set both cross-domain tracking and referral exclusion, but if you have only one domain but has an external serving site like PayPal then set only the referral exclusion (because you actually don't track PayPal in your GA report to pass a client-id to it.)

Finding Answers: Funnel Tracking:

? Goal flow has a lot of potentials that are not available to funnel visualization like segments, and (-2, -1, 1, 2) steps, another important feature is that it can collect data retro; it goes back even before the date of setting it and presents the data of that date, whether funnel visualization report only collects data from the moment you set it up.

? When you set a funnel goal, it will fire as soon as you go to the destination page, no matter you followed the whole path or not. It will fire in all reports, including "goal completion" reports except (funnel visualization reports and goal flow reports)

? Other ways to create a funnel rather than setting goals with steps/funnel is to create a custom report or make a "sequence" using segments.

? If you didn't turn on the button "requires step" that is next to the first step in the funnel goal setting, then GA will "Back-fill" the first step if some user started his journey from a middle step of the funnel and continued to the destination.

Finding Answers: Segments – Part 1:

Segments vs filters:

? Filter is a temporary segment that you can create whenever you want then get rid of it after that without deleting or altering any of your data.

? Whether segments will exclude data forever from your view.

? A cool thing about segments is that you can use the option "build an audience" to build a google ads audience, connect it to google ads, and set up retargeting.

Finding Answers: Segments – Part 2:

? With segments you can choose from built-in or custom segments.

? In custom segments there are a lot of ways to make a segment and the most powerful ones are conditions and sequences.

? You can build an audience using segment, edit it, copy it, share it, make it visible on other views or the only current one.

? Sequence are conditions in a certain order,

Finding Answers: Custom Reports:

? Custom reports are user-specific, which means, it only shows up to the user who created them. So even if you give access to some person to the view or even make someone an admin, he won't be able to see these reports.

? So if you want someone else to see the custom report, you can share it with him from the "share" option.

Tips & Tricks: Dashboards:

? Dashboards are a collection of widgets, so each dashboard is designed to answer a specific question.

? On Google merchandise store you have a "read & analyze" role, the lowest permission in GA, but still can create a dashboard.

? Dashboard is also a "user-level" feature, so only you can see it.

? In widget settings, a "standard report" is the "historic report".

? You can add a filter to the metric displayed on the dashboard.

? You can also create a segment on the level of all widgets

? You can add a link on the widget to some report so you can deep dive into the data that is displayed on the widget.

? Google data studio is made for dashboards but it's a hundred times better than GA. Its visualization is better, and easier to slice and dice data.

Tips & Tricks: Saved Reports & Alerts:

Saved reports:

? You can get into some historical report then add the second dimension, set goal, add an advanced filter, sort by some metric, etc. Till you end with a customized report that answers your questions, now you can save it to access it later with no wasted time. just click save the report and name it.

? You can apply segments to these saved reports, rename, view, or delete them.

Create alert:

? You can set an alert to some events like; 404 pages, or anything so you get notified with them either on phone or email, daily or weekly or whatever.

? When you think about creating alerts think about traffic and results.

Tips & Tricks: Channels:

? Channel is the category of traffic.

? Hierarchy is: Channel > Source > Medium > Term > Content

? Some channels of traffic according to GA (Organic, social, paid, email, referral, direct, other)

? With the "read-only" role (like when you are using Google merchandise store) you can't access the channel settings nor create a new channel.

? The "channels groupings" on the view settings enable you to create a new channel, but it will be available to everyone.

? If you want to create a new channel grouping that is available only to you, you can use the "custom channel grouping" section.

? If you create a new "channel grouping" it will affect the display of the historical data but it takes 24 hours for the data to fully populate in the historical reports. And the same applies to modifying an existing channel grouping that you built.

? Whether if you modified the "default channel grouping", the modifications will affect only, the new data, but the historical data will stay as it is.

? Another place that your channels appear in (other than under acquisition -> all traffic) is with the "assisted conversions" under "multi-channel funnels"

? When you are creating your new channel, the different channels you sort work the same way as filters; the channel on the top captures the traffic that matches it first, then if it does not fit, it passes down to the next one and so on. So the order is important.

So the kind of channel grouping that is available:

? "default channel grouping"

? "MCF default channel grouping" _available only in the MCF -> assisted conversions report_

? "Customized channel grouping", that you build under channel settings

? "custom channel grouping", that you build to be visible for yourself only.

Tips & Tricks: Multi-Channel Funnel Reports:

Standard reports vs MCF reports:

? The way standard reports attribute "goals completion" to different sources: is through the "last non-direct traffic source" before the goal completion.

? So if a user went multiple times to the website, but the last time was through "direct/none" then the goal completion will be attributed to the session before the last one. If he is a first time user, so there are no previous sessions; then it will show up as a direct/none)

? But the reports under MCF are literal; if the source is "direct" then it will show up "direct" if the medium is "none" it will show up "none", they don't try to outthink or second guess, like other reports.

Assisted-conversions report:

? It's the most used report in MCF reports.

? "Assisted/last click" metric: shows the % of the channel being an assistant vs being a converter

?The closer to 0 -> converter.

?closer to 1 -> both assistant and converter,

?exceeding 1 -> converter.

Top conversion paths report:

? It is fun to play with but with so little useful info to take action upon.

? You don't come here to see what is happening but to check if what you think is happening is actually happening. So you can look for a pattern that supports your hypothesis.

? Or maybe you started a retargeting campaign for your email list so you expect to see a common pattern of paid ads being converters with email on the previous step.

? You can get value from this report by looking for specific things; "for this partner "xxx/referral" How his efforts as a referral are doing? Is he a converter with a short path before? An assistant in the first, second, third layer of the path? How many layers is his path before and after?

? Time lag; is the amount of time it took to get a transaction done.

? Path length; is how many touches it took to complete the transaction.

Tips & Tricks; Attribution:

? The "attribution window" is how much time is set for the source or the medium or channel to claim a credit. It's something to set in GA, not in the source.

? The default "attribution window" for GA is 6 months.

? Attribution models are trying to solve the multi-touch attribution problem.

? Attribution in GA is in the "model comparison tool" report.

? You can create your own attribution model

? When using the comparison feature inside the model comparison tool; you should have a hypothesis or certain info you want to check forehand or you will get again into a rabbit hole.

? One great way for using the "model comparison tool" is to compare the first interaction model to the middle interaction model to the last interaction model.

Main attribution models:

? Last interaction.

? First interaction

? In-between (position-based, liner, Time decay)

To be continued the next week, Stay tuned!