Welcome to our new Analytics Roundup! We’ll be covering recent updates in the Google analytics stack (GA4, GTM, Looker (Studio), BigQuery, CoLab), with a few forays into related analytics technologies. If you have any analytics insights to share, resources, comments, or questions, please let us know in the comments section at the bottom. And for more recent developments in SEO & paid media, check out our monthly industry update blog.
Useful GA4 updates
- The introduction of saved comparisons doesn’t quite replace the power of custom segments in Universal Analytics, but it is a major step forward. You can save a comparison and use it like a segment – and despite the name “comparison”, you can deselect the ‘All Users’ comparison to use a single comparison like a filter. A nice thing about comparisons is that they persist from report to report, whereas you have to recreate filters each time you change reports. Pro tip: you can also create a comparison based on audience name if you want to segment by behavior sequences or other advanced logic. I describe the technique in this video. The UX has changed a bit since I created the video, but the concept is the same.
- You can now export GA4 reports to Google Sheets. This is not the sexiest of features, but for those of us who are working in GA4 every day, it’s a welcome addition.
Where Did the GA4 Conversions Report Go?
For some reason, GA4 sort of got rid of the Engagement > Conversions report when they switched the name of Conversions to Key events. This article describes how to recreate it as an Exploration: How to Create Key Events Report in GA4. I personally like the flexibility of the Exploration a bit better than the old report.
GA4 User Acquisition vs. Traffic Acquisition
If you are fuzzy on the distinction between the User acquisition and Traffic acquisition reports (and how could you not be – it’s confusing!), check out this short video: User Acquisition vs Traffic Acquisition reports in GA4 from Analytics Mania. And if that leaves you wanting more, here is a comprehensive explication of attribution in GA4 from Search Engine Land, Your guide to Google Analytics 4 attribution
Chart Titles in Looker Studio
Looker Studio charts and tables now have formattable titles. Ordinarily, I would say that something like this is too boring to mention, but there are two reasons why this is a big deal:
- Tweaking layouts in Looker Studio is a major pain when two elements overlap. If you use a text field to add a title to a visualization, you will almost always have to overlap due to spacing of chart headers.
- Editing a Looker Studio report can be mind-numbingly slow. Fewer elements on a page means less editing.
I wish we had more control over title padding and margins, but it is a welcome addition.
On the subject of Looker Studio slowness, Dominic Woodman put together this helpful guide: Why is my Looker Studio dashboard slow and how can I make it faster? Also a shoutout to Dana DiTomaso, who recommends pausing updates and disabling unnecessary Chrome extensions to make Looker Studio faster. The Chrome extension thing never would have occurred to me, but it really makes a difference!
AI and Data Analysis
We’ve been experimenting with improving our data analysis workflow with AI and have found several tools to be quite useful.
- With Data Analysis in ChatGPT, you can upload data files and ask vague or specific exploratory questions. You can also ask it to clean and transform the data, and to create visualizations. Where applicable, it will do the work but also output the python code it used to generate a result. It’s kind of like having a python wizard sitting next to you, minus the attitude. You need a ChatGPT Plus subscription to use it. At $20/month, it’s also a lot more affordable than a python wizard.
How has this improved our workflow? Our goal when we are doing analysis is to develop repeatable processes that deliver insights on an ongoing basis. One tool we use for this is Jupyter notebooks. ChatGPT Data Analysis speeds up the process of building a notebook by at least 2X. - Google BigQuery recently introduced Data Canvas, an AI-assisted tool for exploring datasets and building SQL queries and visualizations. It’s a bit clunky and doesn’t handle nuanced requests as well as ChatGPT, but it’s a brand-new product and we expect it to get better. I would classify its AI as a mid-level SQL programmer, and that might be a bit charitable. But from a workflow automation standpoint, it’s actually a lot better than ChatGPT. You can essentially work iteratively with its AI – the code it produces is editable and executable right in the interface. It can also be exported to a Jupyter notebook with a click of a button.
How has this improved our workflow? While ChatGPT Data Analysis is like having an on-hand python expert, Data Canvas is kind of like a SQL sous chef. It handles the boring stuff, but we still pull it all together. I would say it also speeds SQL-heavy analysis up by at least 2X.
Consent Mode v2 is a Google Thing
I’ve encountered a bit of misinformation and confusion about Google Consent Mode. Google Consent Mode is a set of features that facilitates communication between a consent management platform and Google tags. If you use Google Tag Manager, you can also configure non-Google tags to work with Consent Mode, but it doesn’t happen automatically. A few key points:
- Implementing Consent Mode does not mean you are in compliance with privacy regulations – only a privacy expert can assure that.
- You can be fully compliant with privacy regulations without implementing Consent Mode.
- Non-Google tags need to be configured for Consent Mode.
- There are certain user-tracking features in Google Ads that require Consent Mode to work in areas covered under GDPR. These include remarketing, audience targeting and website conversion tracking.
- GA4 can model behavior for website visitors who have declined consent if Consent Mode is enabled. There are thresholds that need to be met for this to work, but it is really powerful when it does!
Net/net, I think it is worth implementing Consent Mode, especially if you use Google Tag Manager and/or use Google Ads, but it is a matter of convenience, not a necessity for privacy compliance. This video by Benjamin Mangold does a good job of explaining the ins-and-outs: Consent Mode v2: What you need to know and setup with Cookiebot
The video is a bit of a Cookiebot infomercial, but it’s still very instructive. I’m also an (unpaid) fan of Cookiebot.