Dr. Mario's 2nd 🧠

đź“– Learning Google Analytics

4 min read

Why did I want to read it?

I don’t know. Sometimes I like to treat myself badly.

What did I get out of it?

Very little. Attempts to be a ton of things, but falls flat on almost anything.

The New Google Analytics 4

Everything is an event

GA4’s data model follows a customizable, event-only structure (Location 159)

introduction of GTM Server Side (SS), the configurations possible can also include direct integrations with Google Cloud and backend systems along with modifications of the HTTP call’s requests and responses, giving you the ultimate flexibility. (Location 205)

GA4 events can be modified after they have been sent. This lets you correct tracking errors or standardize events (“sale” versus “transaction”) without needing to modify the tracking scripts—much (Location 235)

There are essentially three ways to configure capturing data from websites: gtag.js, analytics.js, or Google Tag Manager (GTM). In almost all cases, I would recommend implementing them through GTM, (Location 194)

Privacy

OMG, isn´t this extremely black-boxed?

Google Consent Mode is available to remove cookies and their stored personal identifiers so they are not available to Google Analytics until a user gives that consent. However, nonpersonal data can still be useful, and GA4 offers a way to model what your data sessions and conversions would look like if 100% of your users consent to giving their data. (p. 19)

Note that the GDPR requires consent for both first and third party cookies. In other words: no consent means no individual tracking for anonymous users: I could still track events on my own but every session would be a new user for me (I won’t be able to see sessions per user).

Be mindful here to respect user privacy choices. If you are adding information to a specific user, then laws such as the EU General Data Protection Regulation (GDPR) require that you get consent from the user to collect their data for your stated purpose. (Location 429)

What about a logged-in user? A user that starts as anonymous, produces events, and it later logs in, GA4 is able to associate its previous events to the user (and logging in is GDPR-compliant).

What are the key data sources you need for your work goals? What are your main channels for data activation? What do you wish you could do with your data that you can’t do now? What data do you think you should be able to use but can’t? What technologies are you using for your current data work? What are your key business KPIs? (Location 883)

GA4 can and has closed down accounts in the past where PII data has been collected, so it’s worth taking your time to ensure you don’t carry this risk. (Location 964)

Other

“Never dashboard!” stance, (Location 1177)

For GA4 in particular, Audiences is a data activation channel because those can be exported to your paid media channels or Google Optimize within the Google Marketing Suite. If you can link your data modeling to that activity more directly, you’re much more likely to be able to demonstrate a killer use case with measurable outcomes. (Location 1183)

Data layer

Ideally, the majority of your digital marketing data captured via the website should be pushed to GTM via its dataLayer, but you also have the option of using GTM’s selection of web scraping tools to lift the data off the page without coding updates to populate the dataLayer.

The reason the dataLayer is preferred is that it should be less prone to breakage when unexpected changes occur on your website—GTM’s web scraping selectors will break if the theme or layout of the page changes. The best long-term and robust way to avoid this is to involve the web development team in your data analytics process for updates to the dataLayer, rather than relying on GTM alone. GTM should not be regarded as a way to try to circumnavigate the web development team but to make things easier for them to support your web tracking efforts. (Location 1528)

Graph View

Backlinks