#17 - New AI functions, Data Transfer and other updates in October
Hi there,
First, a warm welcome to all new subscribers who joined since the last newsletter.
October was also busy, both with significant updates in BigQuery (look & feel as well as new features) and with vibrant community discussions and tons of knowledge shared.
Going beyond GA4, Facebook Ads got a good bit of an upgrade, giving more flexibility and addressing some of the irritating features (though some still remain). To celebrate that, three articles now cover it end-to-end:
Last, but not least, my MeasureSummit talk on how to build a data warehouse in BigQuery by processing and integrating various data sources (otherwise available with an All-Access Pass) has been made available for our premium subscribers on the videos page.
New AI functions released in BigQuery
Classifying and grouping records has become a whole lot easier by describing the grouping rules in plain English and providing the group labels using the AI.CLASSIFY function, or AI.IF if you only have two classes (binary classification). It is also possible to quantitatively score records against your given criteria using AI.SCORE. Various people have posted examples since on how to use them, like here and here.
UI changes
A relatively big change took effect that changes how we can navigate among objects and assets in BigQuery (the "new Explorer pane"). Important to note that this is where the Job history has moved, and you can't find it anymore on the bottom.
The new Explorer pane, with the ability to switch back to the old Classic Explorer.
When using the new Explorer, you can use the middle part of the screen to navigate among objects and see more details at once. Sometimes it's handier to use the old view where you can expand elements of the object tree: fortunately you can switch back to that view using the middle icon on top, so we really get the best of both worlds here. If you use Repositories, they're available through the right icon.
Another smaller change (or maybe it had always been there, I just hadn't noticed it) was adding a new icon to list the most recent tabs previously opened. Further icons there include Shortcuts (handy!), Gemini (haven't tried) and going full screen.
"Recent tabs" and other icons unlockig convenience features on the top right.
GA4 Data Transfer update
Earlier this year Google released the Data Transfer for GA4, mainly to address the gap left by the fact that the native event export doesn't allow backfilling data. (One crazy person had posted earlier this year that there wasa backfill for the native export, but that turned out to be an April's fool prank.) So anyway, there are some niche use cases where pulling GA4 data through the API is the right thing to do. The early version of the transfer was absolutely good for nothing (prove me wrong), however the new update gives us the flexibility to pick the dimensions and metrics we want (within the boundaries of combinations the API allows to pull in one go). There is a new(?) website with a nice birds-eye view for all the dimensions and metrics, or you can use the good old GA4 Dimensions and Metrics Explorer (handy to try the payload before you set up your transfer).
The new Custom configuration option in the GA4 BigQuery Data Transfer
As mentioned further above, the Facebook transfer was also updated, and there were new transfers released as well, namely Paypal and Stripe.
Some other BigQuery updates
The stream of updates doesn't end here, consider checking out the following:
Early access to conversational analytics. They may have closed down sign-ups for now as I can't find the link to the feature β feel free to drop it in the comments. It is also possible to experiment with the MCP (we've done it at MarketingLens, just haven't had the time to post it); in any case, there will be a webinar later this month to consider, about learnings and takeaways.
Places Insights is now generally available, allowing you to join your data to Google's vast Maps Platform data with business information in a "data clean room" sort of manner β you cannot query the raw data itself, but you can join to your own data and aggregate.
I never got around to sending out any MeasureSummit summaries, but there was one talk that really stood out both in the choice of topic and its delivery: Mobile App Analytics principles by Michal NovΓ‘Δek. I've never seen anyone explain app analytics so smoothly. Fortunately he continues to host webinars, the next one to happen on SKAN next Wednesday.
That's it for now. Thanks for reading and happy querying β I knowingly omitted some community updates because there were just so many... if there are any that you think should be added, please add them in the comments!
#17 - New AI functions, Data Transfer and other updates in October
β Balazs Vajna
#17 - New AI functions, Data Transfer and other updates in October
Hi there,
First, a warm welcome to all new subscribers who joined since the last newsletter.
October was also busy, both with significant updates in BigQuery (look & feel as well as new features) and with vibrant community discussions and tons of knowledge shared.
New on GA4BigQuery
Two new GA4 articles were published, on the analysis of "supporting" products that frequently co-occur with other products (kind of a "prelude" to a market basket analysis using just SQL) and identifying products that drive visitors and transactions on your site.
Going beyond GA4, Facebook Ads got a good bit of an upgrade, giving more flexibility and addressing some of the irritating features (though some still remain). To celebrate that, three articles now cover it end-to-end:
Many companies make a mistake by getting started with BigQuery in Sandbox mode first, and then later finding that their GA4 export data is disappearing β here is a guide on how to avoid data loss after switching off Sandbox Mode.
Last, but not least, my MeasureSummit talk on how to build a data warehouse in BigQuery by processing and integrating various data sources (otherwise available with an All-Access Pass) has been made available for our premium subscribers on the videos page.
New AI functions released in BigQuery
Classifying and grouping records has become a whole lot easier by describing the grouping rules in plain English and providing the group labels using the
AI.CLASSIFYfunction, orAI.IFif you only have two classes (binary classification). It is also possible to quantitatively score records against your given criteria usingAI.SCORE. Various people have posted examples since on how to use them, like here and here.UI changes
A relatively big change took effect that changes how we can navigate among objects and assets in BigQuery (the "new Explorer pane"). Important to note that this is where the Job history has moved, and you can't find it anymore on the bottom.
When using the new Explorer, you can use the middle part of the screen to navigate among objects and see more details at once. Sometimes it's handier to use the old view where you can expand elements of the object tree: fortunately you can switch back to that view using the middle icon on top, so we really get the best of both worlds here. If you use Repositories, they're available through the right icon.
Another smaller change (or maybe it had always been there, I just hadn't noticed it) was adding a new icon to list the most recent tabs previously opened. Further icons there include Shortcuts (handy!), Gemini (haven't tried) and going full screen.
GA4 Data Transfer update
Earlier this year Google released the Data Transfer for GA4, mainly to address the gap left by the fact that the native event export doesn't allow backfilling data. (One crazy person had posted earlier this year that there was a backfill for the native export, but that turned out to be an April's fool prank.) So anyway, there are some niche use cases where pulling GA4 data through the API is the right thing to do. The early version of the transfer was absolutely good for nothing (prove me wrong), however the new update gives us the flexibility to pick the dimensions and metrics we want (within the boundaries of combinations the API allows to pull in one go). There is a new(?) website with a nice birds-eye view for all the dimensions and metrics, or you can use the good old GA4 Dimensions and Metrics Explorer (handy to try the payload before you set up your transfer).
As mentioned further above, the Facebook transfer was also updated, and there were new transfers released as well, namely Paypal and Stripe.
Some other BigQuery updates
The stream of updates doesn't end here, consider checking out the following:
Community posts and goodies around BigQuery
Other relevant blog posts from the community
That's it for now. Thanks for reading and happy querying β I knowingly omitted some community updates because there were just so many... if there are any that you think should be added, please add them in the comments!
Best regards,
Balazs