Measured Opinions #3: What to know before you start with GA4
This week Dan and Dara discuss the four things analytics practitioners should consider before rolling out Google Analytics 4 (GA4). Spoilers, it’s all new… give it a week and this will probably be outdated too!
Dan is still plodding along with the house purchase and Dara promises to share some photos of the company social (if we can remember who took them).
Leave a rating and review in the places one leaves ratings and reviews, or suggest a new topic by emailing Dan and Dara at hello@measurelab.co.uk.
Transcript
Dara: [00:00:17] Hello, and thanks for joining us in The Measure Pod, a podcast for analytics enthusiasts where we try and make some sense of some hot and tricky topics in the analytics industry. And have a little bit of fun along the way. I’m Dara MD at Measurelab joined as always by Measurelab’s longest serving analytics consultant, Dan.
Dan: [00:00:37] Hey, how’s it going?
Dara: [00:00:38] Good thanks, how about you? What have you been up to this week?
Dan: [00:00:40] So this week I’ve been exploring a new feature that Data Studio rolled out which is to enable the GA4 tracking inside of its pages. So it’s always had the Universal Analytics feature and they’ve rolled out a version to enable GA4 tracking. I’ve been playing around with all of the data that we can get, including the enhanced measurement events. Now I’m wondering if we can track scroll depth and engagement rates and things like that inside of dashboards, not just the old classic page view. Just having a little play for now and nothing really concrete, but maybe I’ll report back when I’ve got something to share. And what about you, Dara?
Dara: [00:01:11] So I, I’m not gonna take credit for this, but I’ve been involved internally in us collating all of our own knowledge, our learning material. We’re going to be migrating that to confluence. So I’m working with people internally and then the plan longer term is that some of that content will actually get published externally as well. So it’s really exciting, but as I said, I’m not going to take the credit for the hard work I’m just involved.
Dan: [00:01:37] You can take the credit. No one else is here.
Dara: [00:01:38] True, okay. I’ll take all the credits. Okay, let’s get down to it Dan, what are we talking about this week?
Dan: [00:01:44] So this week we going to be addressing some feedback we got from the first episode we did around “Why should I care about GA4?”. We spoke to a couple of people that maybe we convinced them, now they’re keen to get on with GA4, where should I start? So the title of this week is “What to know before you start with GA4”. I’ve managed to boil it down into four points. So maybe we can just go through those.
Dara: [00:02:03] Yep. Sounds good.
Dan: [00:02:04] So the first point is really thinking about something we touched on a bit in the first episode. It happens to be called Google Analytics and it has the same logo and you log into the same user interface, but it is a different tool. And what I mean by that is this is not an upgrade from Universal Analytics. It is a brand new product. It is a new implementation. So behind the scenes, we’ve talked about it before, but this is a Firebase adaptation. But the biggest thing is that because it’s not an upgrade from Universal Analytics, there is no historical data that’s transferred across. Unlike what we had when we moved from Classic Analytics to Universal Analytics, you could even send both sets of data into the same property and it’ll just take it all and do what it needs to do with it. This is a fresh start. There is no data moved over from Universal Analytics. So a big deal really, try to think of it more as a brand new product, rather than an upgrade.
Dara: [00:02:50] This is a big deal Dan, your right. It’s huge in fact, because as you mentioned, with previous upgrades of GA, the data has remained consistent. Even if there’s been some changes in the data collection, it’s been relatively minor. There is a way around it using BigQuery if you’re at 360 GA customer then you’ll have the native BigQuery integration. Otherwise you have to either use the API or use a pipeline tool like Fivetran or Funnel or Stitch. But it’s not obviously as simple and as straightforward as previous changes because this is a brand new tool. The data collection does start from implementation. There is no import of historical Universal data.
Dan: [00:03:32] Yeah, exactly. So you can always fudge that over in BigQuery and kind of merge the two different tables together and make your dashboard seem like there’s continuity. But yeah, fundamentally, if you were to log in or give someone else access to GA4, they won’t have that year on year comparison that month to month comparison. So moving onto the second point, it basically takes a rethink of all of the tracking. So not only is this a new product, and we said that this is not an upgrade from Universal. What comes with that is that there’s new tags to implement. So in GTM there’s two new tag types. If you’re doing the hard-coded stuff, it’s a Gtag implementation. But it’s brand new tags, we have to implement new things. And with that comes new taxonomies new event structures. So the way to think about it is from Universe Analytics, if I was to track a button and click on my website, then I would be thinking about the idea of an event category, event action, event label and even event values. So all of my data architecture is aggregated into those three layers of category, action, label. And it’s always been that way, right? Whereas now with GA4 has given us way more freedom. But what that means is that we’re kind of stuck because we have to rethink everything that we’ve done that’s been neatly organized into this little pyramid system. Now we have this opportunity to do more. It just means we literally have to go through event by event and decide what we’re going to call it. We’re taking a hierarchical structure of category, action, label and we turn them to a very flat structure of event name and parameter one, parameter two, parameter three, parameter four, et cetera.
Dara: [00:04:57] I see this forced rethink as potentially a blessing in disguise because over the years, obviously lots of tracking potentially gets layered that can be very confusing event taxonomies. And it can be very unclear which events are reliable, which ones aren’t, which are historical, which ones have been updated, which ones haven’t. So this might actually be forcing people to go back to the drawing board and rethink the measurement framework and actually maybe see what’s fit for purpose and what isn’t. And it could be a really good opportunity to improve the tracking overall.
Dan: [00:05:30] Yeah, exactly. Get rid of those random events that are implemented five years ago. And finally, finally, finally get to turn them off. A nice little game you can play just in case you’re unsure if any of the events is still being used, is just pause them. Pause them for a week or rename them even, just in case you don’t want to lose the data. And see if anyone notices, see if it’s going to break any dashboards. I mean, we’re talking about a tool that’s been out for many, many years, and there could be dashboards from before you even joined the company that are based on events that you didn’t know were even set up. It’s gonna take a while. It takes some effort, but as you said, it’s a really good opportunity to have that fresh start, to make a clean break from an old way of doing something that might not even been your doing. And it’s a good way to do it right this time. If it wasn’t the first time. So the next point is a biggie for people that are well into Universal Analytics. And that is that there are no longer views in GA4. And it might sound like just one feature that I’ve pulled out on its own. But what that basically means is because there’s no views anymore, all of our reporting is done at a property level. So we think about all those multi-national websites we’ve worked with that have got French, UK, German, American different websites, and it was very easy to roll out the same GA tracking and then just cut the data into different views so that we can distribute access for example. But now that’s not possible. All of the data is collected into one pot, one property, and that is where you do your reporting. It has a side effect which is something we’re starting to butt up against. And that is that the time zone and the currency is set at the property level too in GA4. So what we’re now saying in GA4 is that all data is reported at the same level with the same currency on the same time zone. It makes it really hard now for these websites that were using views or dependent on using views to aggregate and divvy up access to data. Now there’s just not really a clear way of doing this. So I’m sure there’ll be better ways to approach this over time. Maybe Google comes out with their own solutions to this or viable paths to go down to replicate what we had in Universal. But right now, there is not one answer to say, if you had views, you should do this. It is more of a rethink if we need multiple properties or if we need to do some other method to get to a point where we have that governance back.
Dara: [00:07:40] So previously Dan, with Universal, if we were stuck for some reason with just a single view, we could use segments to slice and dice the reporting data in different ways. Can we do that with GA4?
Dan: [00:07:52] Yeah, that’s really interesting you mentioned segments because they do and do not exist in GA4. They don’t exist in the reports workspace, so when I log in and I browse that left nav of the reports that we can now customize. We can’t use segments in there, but we do get something called audiences, which technically did exist in Universal, but that’s more of a permanent, non backwards looking aggregation. So if I create an audience looking just at UK pages for example, that will only be available for me to use going forwards. You almost need to know before time and get those set up as soon as possible.
Dara: [00:08:22] And also segments would be the second choice in compared to filtered views. So fingers crossed Google are going to introduce something that is equivalent to filter views, even if it takes a slightly different shape or form.
Dan: [00:08:37] I hope so, fingers crossed. I’m really struggling at the moment, in some ways with GA4 to just filter out data I don’t want to keep permanently across anything. They’ve got that automatic bot filtering checkbox that they had in Universal which is great, but it’s not identifying enough. As well, when you link Google Ads to your property in Google Analytics 4 it automatically syncs all the audiences and conversions. So let’s say I go back to that multinational retailer, where I’ve got the UK, American and German websites all rolled into one property. And let’s say they all have their own independent Google Ads accounts. Every single one of those Google Ads accounts will have every conversion from all countries synced, let alone with all the audiences sinked as well. You’re having to then share the, the limits of audiences between all those countries so whoever gets in there first gets the most. There’s a couple of tweaks they need to make I think before some of these more convoluted Universal Analytics setups can fully migrate across. I still would recommend to do it. I still would recommend to get GA4 setup and just get data collecting. Think of it more as a training exercise for people that are using GA to get familiar with the tool. But I can imagine at some point in the future that the account structure might have to change so that you get to a point where you’d be happy or at least non panicking if you had to turn Universal Analytics off.
Dara: [00:09:49] If there are limitations or drawbacks on how you link up with Ads, then that’s probably the single reason why this is likely to change.
Dan: [00:09:57] Yeah, it directly affects Google is pocket. I think we can be pretty safe to assume that they’re going to be looking into this as soon as possible. I think that kind of ties neatly into the fourth point, which is that it’s new. It’s simple as that, it’s just a new product. You will find as you use GA4 lots of things are still labelled as beta, things change almost every other week. They’re releasing new features pretty quickly, which is fantastic, but everything is subject to change. For example, the reporting API is in beta. It recently was an alpha, and now it’s going into beta, which is a step up, but it still doesn’t even plug into the Google Sheets connector. So for anyone that’s doing any kind of reporting using the API through Google Sheets, you can’t do it with GA4 just yet. So if you’ve got reports based on that, then we’re kind of out of luck for the time being. And I know that’s a keen favorite tool of yours Dara.
Dara: [00:10:43] It is, I’m eagerly looking forward to that.
Dan: [00:10:46] Also the fact that it’s new just means that it’s unknown territory, right? And just the fact of it being new and unknown means that we’re all figuring it out together at the same time. So basically, best practices just don’t exist, or if they do exist, they’re very quickly outdated and will change all the time. We’ve been playing with this for quite some time, comparatively compared to a lot of other people, but that’s only been a year and every month I think is almost something brand new to learn. So although we pride ourselves with knowing as much as we possibly can with GA4, it’s brand new, we’re still trying to figure out, and we might have some best practice that Google might roll out a feature for the replaces or changes that opinion of the next day. And as the industry gets more familiar with it, as we start using it more, as these use cases, these niche or edge use cases come up and get addressed, I think we’ll kind of settle down in terms of a series of best practices. But for the time being, I think it’s just being comfortable with knowing that you’re doing the best we can with what knowledge we know at the moment.
Dara: [00:11:43] Brilliant, thank you Dan. As always, I get the job of concluding and that I think it’s pretty easy this time round. I can sum it up in two words, it’s new. So new technology, it means a fresh implementation, which means an opportunity to rethink your data collection. One of the consequences of that, obviously, as you mentioned, is you don’t automatically get your historic Universal data. So if that’s a requirement, then you need to come up with a work around using BigQuery. The structure is different with the most noticeable example being the lack of views so that the idea of a filtered view doesn’t exist currently in GA4. And probably most importantly, and tying back to something that we mentioned in our first episode, is that it’s continually being worked on. So all of the dev resource on Google’s side for the analytics product is now going into GA4. So we’re going to continue to see a change. And as you rightly mentioned, best practices will therefore change over time as it continues to evolve. Is that a fair summary Dan?
Dan: [00:12:44] Yeah, it’s new, right? A new puzzle for us to solve.
Dara: [00:12:48] It keeps us on our toes. Speaking of what’s new Dan, what’s been happening outside of work for you lately?
Dan: [00:12:54] Well, a big milestone actually in my life. I’ve just had an offer accepted on my first house. So I’m right at the start of this journey and I’ve been told it can be quite a stressful process, but right now, touch wood, it’s been quite a smooth process. We’ll see.
Dara: [00:13:08] So no better way to wind down than adding lots of extra stress outside of work.
Dan: [00:13:13] Yeah. What, what’s an extra couple of plates to keep spinning.
Dara: [00:13:16] And it’s a good stress to have it’s really exciting news for you. So I’m very happy for you.
Dan: [00:13:21] Oh, cheers mate. Yeah, fun, stressful and exciting overall. So what have you been doing to wind down, what’s new with you?
Dara: [00:13:28] Well, as you will know cause you were there, we had our first company social in real life for quite a while given everything that’s been going on with lockdowns and COVID et cetera. So we had a get together on Friday which was really, really nice, really great to get to see people in real life. A few of which I hadn’t met in real life before. So it was really good. Fun. Few drinks were hard. All very sensible though.
Dan: [00:13:54] A few drinks were had indeed, yeah. It was a wicked evening, not too crazy. And it was nice to see new faces that I’ve only ever seen with a box around them on a screen. So to get them IRL was wicked.
Dara: [00:14:06] And the weather was amazing. I think we had all four seasons in the space of four hours.
Dan: [00:14:11] Yeah, it’s almost like we were catching up for lost time.
Dara: [00:14:14] All right, I think that’s a wrap for this week Dan. As always, you can find out more about us over at measurelab.co.uk, or you can email hello@measurelab.co.uk if you have any questions or if you’d like to suggest a topic for us to discuss here at The Measure Pod. Otherwise, join us next time for more analytics chit-chat, I’ve been Dara joined by Dan. So it’s by for me.
Dan: [00:14:35] And by, for me.
Dara: [00:14:36] See you next time.
I think you said, how are you or something and I went I am well, I, there was, there was something I shouted at anyway and I thought I might redo that, but I’m sure I’m sure It’ll be okay.
Daniel Perry-Reed
Subscribe to our newsletter:
Further reading
Customer Data Platforms (CDPs)
Grand Designs: “The Warehouse to CDP conversion”
