Drill to Detail - Drill to Detail Ep.97 'The Role of GA4 in the Future MarTech Stack' with Special Guest Daniel Perry-Reed

Episode Date: May 12, 2022

We’re joined on this episode of Drill to Detail by fellow Brightonian Daniel Perry-Reed from Measurelab, talking with Mark about the new Google Analytics 4 release and its role within the future Marketing Technology stack.From the wild west of data collection to conversion modellingThe sun is setting on Universal AnalyticsMeasureFest April 2022: The role of GA in the future MarTech stackEvent-Based Analytics (and BigQuery Export) comes to Google Analytics 4; How Does It Work… and What’s the Catch?The Measurepod

So hello and welcome to another episode of Drill to Detail and I'm your pleased to be joined by Dan Perry-Reed, Innovation Lead at MeasureLab. So Dan, nice to have you on the show and yeah, it's good to have you here. Yeah, thanks for inviting me, Mark. Yeah, it's good to be here. So Dan, tell us about how we know each other really, because I think we've known each other for a while now and you're also a podcaster as well and you actually work just down the road from me. So tell us about what you do and the role you do at MeasureLab. Yeah of course well we know each other because Brighton is a very small town and there's a couple of marketing firms and agencies and analytics companies that are all kind of existing around the sort of the Brighton to

London remit and yeah we're both in Brighton well we're just outside of Brighton in Lewis so MeasureLab we've been here for a little while just like you Mark and I think we're both in bryan uh well we're just outside of bryan and lewis so measure lab we've been here for a little while uh just like you mark and i think we just attend the same events we do the same meetups the same uh talks and yeah i think it's hard to not bump into each other in a city so small as brighton um but yeah it's all we i mean at measure lab we are an analytics consultancy and we just do analytics focusing on the Google stack, Google Analytics, Tag Manager, Data Studio, dipping into things like the GCP and other peripheral tools. And my role is, as you said, analytics innovation lead,

which is a bit ambiguous, but my job really is to know as much as I can about these products to help our clients apply that into their instances and to turn that into podcast episodes and to videos and to training content really just to help spread the spread the love spread the knowledge of these products okay so i know you do this as well in your podcast but um i'll ask you now how did you get into the industry and how did you end up at measure lab now it's a

really good question and i love asking it myself probably why we adopted it in a in our own podcast as well um but yeah because there's no there's no one way into analytics is there and uh my journey sounds actually uh like it is an obvious choice but at the time it was very unobvious so i uh studied maths at college i did a couple of maths and physics a levels and then i decided i wanted to go to university just because i didn't know what i actually wanted to do with my life so just to kind of postpone the inevitable decision of actually having to figure something out i decided to go to university and of course i thought i can do maths maths is easy for me surely it'll be easy for me a bit of a free ride in university so i decided to take on mathematics at the university in sussex which is uh in Brighton. And God, I was definitely in for a shock because

it was hard as hell. But nevertheless, I got through my undergraduate degree and I did a fourth year. I did a master's at the university and randomly I got two emails in my last ever week at university. I got two emails that came through to my student email address. One was for a company called DC Storm, which is an analytics and attribution company, or at least was an analytics and attribution company in the middle of town. And the other one, I honestly couldn't remember what it was for. Um, but either way I replied to both saying, yes, please. I would love to have a job because you know, I'm now at the end of my tether in terms of how many times I can delay the inevitable. Um, and, uh, they both got back to me and one of them offered me an interview

and I got that job and little did I know it, but that was my entry point into marketing analytics and is very much at the deep end in an attribution platform. So I very quickly learned everything about marketing, data, implementation, tech management, attribution modeling and everything. And it was just a proper proper whirlwind in the

deep end but i loved it i i got to play with data in all sorts of different forms i got to talk with loads of different companies and um yeah i spent a couple of years there before moving over to a company called measure lab which i'm at now and uh yeah i've been there for the last six years um doing more of the same actually but just really specializing in you know the google stack and uh very much where where i've invested my my development my knowledge and my my education okay okay so so in case anyone's thinking you know i've just invited somebody i met down the pub um to come on the present come on the uh the podcast um the reason that i i want you to come on was um as most of the intersection or in the worlds that you work in in the worlds

that we work in so so you know we work with uh tools like uh gcp and looker and and and build uh i suppose modern data stack analyst analytic platforms excuse me and and you build uh well you work in the world of kind of google marketing and ga4 and so on. And in particular, I suppose recently there's been quite a convergence in some of the tools and techniques and technologies that we work with. So I've been working on some GA4 projects recently, and I've noticed that GA4 is getting more like some of the tools that I'm more used to using, like Segment and so on there. But in general, I suppose there's the technology um overlap that we've got but also really the sort of the changes that are happening in the industry

um that i want to talk to you about particularly around um i suppose the effects of some of the privacy changes that have been happening recently and i suppose also the um uh or really what is the implication of that really and how we do things like attribution and so on so um that's the kind of context for this but but maybe just to start things off uh dan um tell us what what is what is ga for what is google antics for and and you know why is it a big deal at the moment really just in in sort of general terms that's a really good question um and i always like to ask or try to answer why it's called number four when there was no number two three or one um but but actually it's a a brand new product or the way we can think about

it's a brand new product from google that does a very similar job to universal analytics in the previous versions this time it's built on the firebase technology stack rather than the urgent analytics stack that the Google Analytics came from. And what it is, is a marketing analytics product at its core, it's a marketing product. And it tries to measure the performance of your marketing campaigns to an onsite or on-app conversion. And it tries to assess which ones are working

and which ones aren't. So at its core, it's a very, I don't want to say basic or simple, but it's a very straightforward product that tries to do one thing, one thing very well, which is measure your marketing performance to be called Google Analytics, unlike previous versions or upgrades to the product. This isn't a continuation of the existing data. The data scheme is completely different and everything else. This is a you have to implement and start from scratch. There's no historical data, like I said before, continuing from Universal Analytics. analytics so in a sense this is the biggest shake up that the google analytics technology has had

probably since it slapped the google analytics logo on the urchin analytics product back in you know the early noughties um so yeah it's it's big it's a big deal because it's um it's kind of uh daunting and maybe kind of scary for a lot of people to have to get rid of or not use the historical data which maybe is more of a safety blanket than a real necessity nowadays. But a lot of people are having to walk away from maybe up to 15 years worth of historical data, which they can't take with them. And the other side is that, I suppose we'll dig into this a bit more, Mark, as we go through this conversation, but actually it's the approach that it's taken to the data, especially around its modeling capabilities.

And fundamentally, there's almost less of a focus on it being a data product and more of a focus on it being a marketing product. So there's less focus on it being data for reporting. It's more like Google Optimize. So it's a change of focus on what was a very well understood product. And I think that's the biggest kind of shift that it's gone through in the last couple of weeks, specifically since the announcement, but since it came out in 2020. Okay. Okay. And then beyond that, really, and this is where think it also even if even if our listeners aren't users of ga4 or going to use ga4 some of the changes that are happening i suppose in in in legislation in in in those attitudes about how data is collected and stored they have they're happening at the same

time and they've obviously had some influence on or a lot of influence on the way that GA4 has been put together so we'll cover it in more detail in a moment but at very kind of high level what is the general change in the industry and and how we do things that is is informing how GA4 works? I think it's a couple it's quite multifaceted actually I think there's the the obvious one which is the the legal side and I think there's a lot of new laws coming into effect or even old laws that are being revisited and realizing that things aren't as compliant as we once thought and but another one is kind of cultural or social and i think just the general awareness of like uh tracking and privacy and a technical technological savviness when visiting websites and apps online and just what they're doing behind the scenes and how they're sort of using your data to kind of capitalize on i suppose i think there's just a

cultural shift in how just the average consumer is aware of of what's going on with the data so i think there's a couple of things that it's that it's trying to address or trying to solve for and i think that's that's the kind of approach that google have got with google analytics for trying to solve for these changes in the industry um the um the first one being the the legal changes in different countries and territories having different sort of rules and regulations around how you can track and what you can track there's a cultural shift in terms of people's awareness and willingness to give up access to that kind of data um and i suppose the third prong in this um is has to be the technology changes so things like safari and firefox with itp and etp and att app tracking

transparency from apple but all these things that are happening with the technology regardless of consent regardless of users awareness regardless of the technology you're using are having a quite fundamental uh impact in the tools including GA4. Okay okay so so again another reason I wanted you to come on the show was you you did a great presentation on on this topic at the recent MeasureFest FringeF in Brighton and we're going to use that as a I suppose as a kind of framework to have this discussion really but but just to probably get into that the detail of the conversation you know what maybe what is what is measure fest and um and what were you looking to try and get across in that presentation really just to give it a bit of a kind of a trailer really

sure measure fest is a i suppose the the most local analytics specific uh conference we have and it's uh as you said a fringe uh event to brighton seo so brighton seo is a huge probably internationally known seo conference in brighton but there is the the kind of sister fringe events one of which is a analytics and cro focus one measure fest so it's very much a a good place for people like us or me specifically to go hang out and talk at and just to kind of get with like-minded people so measure fest is a a smaller version of brighton seo the day before um and just specifically for analytics um people and and and talks okay okay so so let's start then really by i think one of the first things maybe to think about. So a lot of people listening to this podcast will not know some of these terms you've been talking about and will not know nearly in as much detail as you about, I suppose, the

history of Google Analytics and Google's kind of marketing technology strategy, really. So let's kind of wind the clock back a little bit to when you mentioned Urchin earlier on. Now, I know what you're talking about, but most people probably, maybe people wouldn't. What was, how did Google's products in this area start out and what was the strategy behind them, do you think, when they first came about? That's a really good question. And like anyone in this industry,

I think we use more acronyms than we can even define ourselves. So thanks for calling that out. I'll try to define everything as we go. But the way that I kind of approached it from my talk at MeasureFest is I realized a lot of people are going to spend a lot of time talking about the technical features and tools and the approaches, especially this whole GA4 migration. So to kind of throw that completely on its head, I decided to talk about the history and the future of Google Analytics 4. So look back in time to kind of predict where it's going in the future. And the way I did it is I used a Marvel MCU analogy as much as I possible. I very much stretched that analogy as much as I can. And so I defined the

kind of first phase of the Google universe, the Google marketing universe as this web domination phase, going back all the way to the year 2000 where they they introduced google adwords where they started to kind of monetize the google search results um but where google analytics kind of shows up in this story is that um at its core google analytics wasn't it never used to be called google analytics it used to be called urchin analytics it was an independent analytics platform called urchin and it was actually founded back in the 90s so it's had a long legacy uh over a number of decades now um but google ended up purchasing urchin analytics with a primary focus really to measure how well their google adwords platform is doing so they needed a tool that measured websites

uh that they could measure their own advertising platform and the success that they're bringing i suppose on the website but actually it's brought is broadened that so google analytics isn't just google adwords analytics otherwise it would have been absorbed into its you know the actual product itself they kept it separate because it is a multi-channel analytics product it doesn't actually matter if you're running google ads or not or adwords as it was back in the day it is a independent tool in its own right that measures your website and everything that comes in and everything that happens on it so um really when thinking about the the kind of phase one or at least the way i approached it in that that presentation the way i thought about it is that it's a web domination phase it's the kind of starting in the year 2000

when they started to monetize google adwords leading up all the way to the release of universal analytics so a number of versions after urchin and google analytics and and the kind of the evolution of the product there so just to kind of bring all of their advertising and measurement solutions into one happy suite as it were one happy family um but yeah over that period of time they they bought a platform called double click which was at the time at least pretty game changing it was the sort of independent uh it was sorry it was the like industry standard way of measuring marketing so they kind of absorbed a lot of products and then they kind of rolled them all in together that's why i kind of thought of it as a web domination phase because they're kind of dominating by absorbing and buying up all of these products and rebranding them as google products um all kind of under the guise of of measuring you know

google ads and you know adsense and double click uh campaigns whatever whatever google product you're advertising through okay okay and so and and that just again for anyone who's not familiar with it there were two i suppose two editions of ga or two sort of versions there's the free one which most people think of as ga and there's the kind of paid for version and which i suppose will you tell us what what was the what did you get with the paid for version that made it worth investing quite a lot of money in really at some points you know what was in that that kind of suite of products all that product really yeah that's a good question and i don't think there's one answer of why why um why you end up paying for google analytics premium as it was or ga360 as it is now known and um in the in this kind of um version of

google analytics at very least there's a very hard cap in terms of how much data you can send in per month so what it does is it raises some of those caps so it basically if you if you've got a lot of data volume then you need the paid for products just to kind of open up access to the data you're collecting um and i suppose a a side effect of that is something called sampling so when you collect a lot of data into google analytics to save some of the resource you know the resources um uh in the cloud it when you run a report in the interface rather than trying to query all of that raw data and give you the answer you need it will take a subset of that and do some some some modeling do some sampling and kind of give you an estimation it's a very very good estimation but nevertheless it's still an estimation and when you go to the premium pay for solution you again those caps get raised so

you have less sample data and more exact data so in a sense if you've got a lot of data and you want access to it accurately you kind of have to have to pay for it okay and so where i think where i mean i was i was always aware of a free ga and everyone everyone installs it on their websites and so on but where i suppose it came into my world was when we had customers who had ga360 and they wanted to bring the data from ga360 into for for example, BigQuery or things like that. And that was actually, I suppose, that was a request and something that was being done by a lot of companies with other technologies as well. So we had customers using, say, Segment or using other, I suppose, event level or certainly detail level sort of tracking tools that allowed us to really track customer behavioral activity across both mobile and web and so on. And that was a kind of world, wasn't it, that is no longer there to some extent,

but where everything could be tracked. And then you had kind of companies building attribution solutions that could track everything and so on. I mean, describe that world really and how that would work and why that's going away, really. Yeah, no, it's a really interesting way of phrasing it. It's that everything's tracked, everything's shared. And I suppose that you can think of it as third-party data ruled supreme, right? So whatever happened on your website or app, every single vendor had a version of that or a copy of that piece of data. And so whether it's Segment, whether it's Google Analytics, whether it's your Facebook pixels or your floodlights, every platform had access to this data, you know, regardless of whether you knew it was happening or not as the consumer.

And then every platform could do their own attribution modeling. They could do their own data storage, or you could pass that into your warehouse or whatever you wanted to do with that data, really, which was great and made a lot of companies a lot of money, right? Especially in the advertising space like Google and Facebook and Amazon and platforms like that. But what actually has started to happen, and we're seeing this probably more so in the last maybe year or two than any other time, is that the legislation and the technology changes that are kind of catching up with us now. So in a sense, what's happening is you can't track everyone doing everything you have to ask for you know explicit opt-in consent you know

whether that's through things like the gdpr or other local regional legislation but also things like cookies you know which most if not all of these platforms are built off of your third-party cookies are blocked in pretty much every browser except for chrome and first-party cookies have got some huge limitations on them in terms of their longevity now so identifying the same person coming back multiple times before they you know convert or make a purchase on your website is becoming harder and harder and harder to do if not impossible so um what's happening now is this the idea of third party data reigning supreme is kind of almost very old-fashioned now and it's all about first-party data so there's one version of your data which the generally the the company

itself would kind of ask your consent to collect and clicked for their own benefit but then sharing that data with third-party tools like google like facebook like other things becomes harder and harder if not impossible to do and i And I think this is where we're seeing the big change or the tide changing on this is rather than everyone having access to all the data, making a lot of money off of it, doing whatever models they want or optimizing whatever ads they happen to be running off the back of it. What we're seeing is the inverse where they're having access to less and less data. And I think this is where something like Google Analytics 4 has been brought in from the Google ecosystem

to try and solve for. I think if I even remember rightly, they even say on their announcement of the Google Analytics 4 platform that it's there to, let me get the quote right, it's privacy-centric by design, so you can rely on analytics even as industry changes like restrictions on cookies and identifiers create gaps in your data so it's a very very specific

problem they're trying to solve for as access to your first party data becomes more limited we are trying to get around it by doing some clever stuff okay so let's again just just for anybody who is is is kind of new to this it to this, let's define a few terms here. So you talk about first-party and third-party data, and first-party cookies, third-party cookies. Can you just, for anyone's benefit, can you just explain what you mean by that, really? And why is one on the way out, but another one of those is still relevant and valuable and so on? Yes, of course yeah so um in a sense all cookies are exactly the same and they're small text files that are saved on your browser um per website a

first party cookie or i suppose the more technical definition is a cookie saved in a first party context is um means that the cookie the text file that i'm saving is per domain per browser so um in a google it's really a really good example is talking about the google analytics cookie it's a random string of uh alphanumeric string which i then it's a random user id basically that identifies me as me but if i go to a different website i've got a different cookie and if i go on a different browser i've got a different cookie so in a sense these tools can't track people. They can only track browsers and cookies at that, which are quite fragile because people can clear cookies. They can go incognito where cookies aren't saved, all these things. So that's the first party cookies. In a sense, they're the

good ones, right? They're the ones that, yeah, not that cookies can be good or bad, I suppose, but they're the ones that a lot of products and technologies rely on. The third-party cookies or cookie saved in a third-party context basically means rather than saving it on your domain against your domain, so let's say we go to the measurelab.co.uk website and it's saved specifically for that, it's saved on the advertiser's domain. What that means is that someone can measure the same person across multiple different websites as the same person, as the same user.

So this is very much that kind of big brother, you know, the kind of CCTV cam monitoring you going across different websites. So this is very much why third party cookies are being unsupported and being deprecated across every modern browser. And people are stopping to use third party cookies. And quite often it gives cookies a bad rep because people think cookies themselves are bad and tracking you they're not but third-party cookies um enable platforms to be able to do that first-party cookies are very um very basic in comparison but they can't track you across different domains they just track you on the one website doing the one thing. Right, right. So if you went to, for example, the, you know, the MeasureLab website and you have GA installed on there, you know,

that would be a first-party cookie that's being set up there that is not being used to track you beyond that site, even though it's actually being kind of run, I suppose, by Google Analytics. It's a first-party sort of cookie. Is that correct? correct yeah that's correct yeah right brilliant okay um so so i suppose there's that happening so so this is so i suppose what they're trying to outlaw is or or legislate against is those things you get when you go to one site and then you go another site and then you get an advert for the thing you looked at on the first site saying you know and it's it so i suppose that's retargeting isn't it so it's that kind of thing that is getting less possible. But what, but I suppose,

even within the world of first party cookies, GA4, you talked about privacy by design, and you talked about opt-in and so on. So, so, you know, let's, let's kind of talk about, I suppose, what GA4 is doing differently to say the previous versions of GA and how that affects um you know what data you collect on somebody so so that part the point you made there about or the quote you had about is privacy by design what does that mean in practice in practice what it actually means um is that well let's actually start with um with how we can collect data so first of all we need consent to collect any data right in in pretty every territory. And even if it's not, um, uh, you know, even if there's no legal requirement to do so that I think there's more, almost like a moral or ethical requirement to do

so nowadays anyway. So, so what if I don't give you consent to, to track me, right? What happens then? And, and, and there is basically just a hole in the data, which means Google analytics and these other platforms uh theoretically can't track you or what you're doing or who you are which makes perfect sense if i don't consent for it it shouldn't happen but what and the reason that's an issue is because if i can't measure the conversion occur i can't measure an roi against the advert that we've been running right spending money on so in terms of if you think about it from maybe Facebook's perspective, um, you're investing money in Facebook, but you can't measure any revenue. So there's no ROI, which means that all of my adverts look like they've

got negative ROIs or basically they're performing badly enough to stop running them, which is a big issue for these advertisers like Google and Facebook, because if every advertiser goes into their platform and sees I'm losing all my money on every ad I'm running, people aren't going to be incentivized to continue spending money in their platform, which obviously hits their bottom lines quite hard. So what Google Analytics have come up with is a way of modeling the difference. So when people don't consent to be in tracked, Google uses the data they can measure and they flesh out the middle bits. So basically they try to model in the unconsented users using their machine learning models and various different, very clever techniques that I'm,

I'm very, not very privy to, of course. So that's, that's what they're trying to solve for. They're basically saying where we don't have access to the real data we're going to do some modeling and fill it in for you so that in a sense you still maintain total data even if it's not 100 accurately tracked it is modeled to kind of fill in the gaps that's not something that even the previous versions of google analytics have have got available you know if you don't consent you don't have the data it's as simple as that whereas google analytics for trying to where it says it's privacy centric by design and uh it can evolve as the industry changes and gaps in your data occur that's what it means it's going to create models to fill in those gaps okay so so when you say so consent so so we've all we've all encountered

those those those kind of i suppose cookie pop-ups you get everywhere right where it says you know track me and so on there um and so presumably then when people install those there is something that then takes that information from the customer from the person who's viewing and they they set consent on there so so so what what is that process that you do and what happened what would happen then if a website didn't have a cookie a cookie sort of like a question on there does it then track by default or how does that kind of work that's a really good it's a really good question and um there's an an awful lot of trust that one puts in websites to have these installed properly as well because i quite quite often in in my line of work you come across um websites that have cookie banners or at least you know there's a pop-up you have to accept or decline um but actually um i i like to think of it as plumbing they haven't plumbed it in to do anything and

it's almost like a vanity project where they've got that you know check we've got our cookie banner let's move on yet it's not actually preventing any cookies from being set or any tracking occurring so um the way that the way that it should work is that um first of all cookie banner isn't tracking so cookies don't do tracking. The code on the website, the product you're implementing, like Google Analytics, does the tracking. But it uses cookies to identify a new and returning user, the sessions, the visits, the bounce rates, those kind of things. So the cookie in itself is harmless. It's the products that you're implementing do the tracking.

So when you're accepting or declining cookies, you need to make sure at least we need to have faith that the website is actually dynamically serving these products like Segment, like Google Analytics, like other products based on the consent levels that have been provided. One would imagine if there's no banner, then you're just being tracked by default without any respecting of consent so um so yeah realistically i mean in in the uk at least in the eu we've got things like um peca so pecr is a legislation that's in place and gdpr kind of expands a bit upon that slightly these are all um probably things that we've come across

or at least gdpr we've probably heard even if you're not even in this industry we've probably heard um but ultimately what the gdpr has done for the marketing and the marketing analytics world is that because you need opt-in consent before you do any of this stuff you can no longer rely on assumed consent or implied consent to do this so now what it means is things like you mentioned before mark around things like retargeting you go visit a website then you go onto instagram you see an ad for the product you've just visited that theoretically or legally shouldn't be possible unless you've given consent for that to happen um and we know that not everyone in not every website ever is going to be compliant with this stuff, but that's what this means.

Okay. And also from my own looking at working with GA4, even if you do track by default, things like IP addresses aren't captured anymore. So I suppose PII data. Maybe just talk to us about what GA4's approach to PII data is and how that differs from before and other tools. Yeah, for sure. So PII, I suppose, actually has a before and other tools. Yeah for sure so PII I suppose

actually has a different definition depending on what country you're in I think that's actually one part of the complexity of doing this kind of thing so personally identifiable information I suppose what PII stands for is any information that could be tied back to you or identify you as a human being from the data itself and there's you know depending on where you are there's different definitions but in Google Analytics specifically within their t's their terms and conditions um you can't track pii data in google analytics so you can't track things like names email addresses or postal addresses um you can't track those things so if you're sending that in you can actually be banned and your data could be deleted your account could be suspended so

there's a lot of um a lot of effort they've put into not allow people to put in that kind of stuff however in again in the uk within gdpr things like the um that first party id that we store in that cookie there's a bunch of gibberish um that it could be considered personally identifiable information now so depending on where you are so where we are mark and that that first party cookie id is pii that is personally identifiable so um what that means is that technically we can't collect that into google analytics unless we have consent and that is the biggest change um again i don't reiterate but at least where we are um and and that's and that's what that's what they're trying to solve for so when you don't give consent, you can't collect the ID. So you still, Google Analytics knows kind of like server logs.

And server logs are, if I have a website, I know how many times my page has been served, right? Because I have to know that otherwise I can't serve you the page. So in a sense, tracking the kind of basics, the counters, the counting elements of it, how many times has this page been viewed? That is always possible with or without consent. It's just, can I tie that to a user? Can I tie that to a session?

Can I tie that to some marketing campaign? These are the things that we start losing, including things like you said, IP address, when we don't have that explicit consent. Okay, so what this is all leading to really is how does this change the world? How does this change the world? How does this change the game when you are building, say, for example, us, we're building an attribution model

for a customer or when we're building a conversion optimization model. So what fundamentally has changed about how you have to do things now, now we are introducing model data in, and we don't necessarily know what each individual person is doing maybe start with the simplest thing of just tracking conversions and the and the kind of conversion optimization yeah of course um so the first thing is is that we might not even know the conversion occurred quite simply um if there's no consent we we can't track anything or at least we can't tie it to that user. And anything to do with attribution modeling or understanding the customer journey, the one commonality we have to have with that is understanding the customer or that it's by the same person or the same device.

And that unfortunately is the bit that we can't track anymore. So the way I kind of think about this is that attribution modeling is still completely possible but the data set that you have to do your attribution modeling on is shrinking so we're not going to have access to every single person every single conversion and the history for that user leading up to that point of conversion so it's just about again the approach and the models and everything else stays exactly the same but the data you actually have available to you uh the volume of that data the touch points that you can attribute across is going to kind of shrink and i can imagine continue to shrink over time right okay and i mean this is are we talking about largely anonymous access here where it's like a b2c site or something where people don't log in i

mean what what is this different when you're talking about business to business sites where people do log in and you do, people do give you, they do identify themselves, for example, is, is that a different kind of conversation really? I think it is. And it has to be, I mean, I think that's actually how Google actually manages to continue serving ads and operating themselves. You know, everyone signed into something within Google, right? So whether it's Gmail or Chrome chrome or or youtube or something like that and that is their logged in process that they can kind of identify you across lots of different uh websites so so yeah i think if you if you're a you're a sas company or if you've got a login port if you're amazon for example even a you know a retailer but you know you have a login portal um as soon as you log in you can put whatever you need behind

your t's and c's you can say okay if you log in you give us consent to store cookies etc etc because in a sense you need you've already got that opt-in consent right um so i think once you're logged in yeah you can you can do everything because it's the same difference as clicking accept all on the cookie banner uh as soon as you log in because it's all tied it's all tied into the same kind of like contract in a way when you sign up for an account so that's why a lot of a lot of at least i've noticed a lot of websites in the last couple of years have moved very much on to kind of pushing you to log in or create an account you know people people don't necessarily want you to check out as a guest on an e-commerce website because they want to identify you and tie this data all together um so yeah that's that's very much um you're quite right to

kind of call that out if you're logged in um this isn't an issue however before you create account or before you log in is still invisible so um if i if i log in you know halfway through a visit onto your website um all of those previous page views all of that you know whatever ad i click to get to your website all of that is still invisible it's only going to start sort of stitching things together from midway through when i can i can i log in or click sign up okay so i'm curious to understand what your clients at measure lab are doing with this kind of situation that's happened i mean think about in some respects this is these changes although it's, although they're being, although they're being kind of sold or communicated as being sticking

it to the man in some respects, you know, this is stopping big businesses tracking us and so on. You can imagine, you know, your average, you know, UK retailer or your small business that is using, that is maybe sort of relying on getting audiences from Facebook and so on. You know, I can imagine they would be quite hit by this, really. So I suppose, what are businesses that previously have done this kind of work with you? What are they doing now to try and sort of stay competitive and have a view on their own customers' behavior? That's a really good question. And you're quite right.

It all comes from a good place. You know, introducing things like the GDPR and other legislation all comes from a really good place of making sure it's more private and more control is given to the consumer browsing the internet. However, what it's actually done is it's pushed people, it's pushed advertisers specifically into the pockets of these big monopolies. So I like to think of them as walled gardens right so you've got the google ecosystem the amazon ecosystem and the facebook ecosystem and because these smaller advertisers can't compete anymore because they have no data to be able to kind of you know help you get the sales that you need to do so what i found with my clients at measure lab is they've probably gone one of two ways uh one way is to lean into the product so if you don't have the the data engineering or the data science resource or the availability

in terms of that skill set in-house or through an agency, quite often you just have to, in a sense, just be aware of what Google Analytics are doing, but just use it anyway. Maybe not make rash decisions off the back of the numbers you think are telling you one thing, but are telling you something else. But leaning into the product. So Google Analytics are quote unquote solving this. They're're modeling out the differences they're doing data-driven attribution they're doing in a sentence there's a black box giving you an answer um the other the other way around is with people that have access or organizations that have access to that um

resource you know the data science the data engineering resource or or ability um is actually kind of pull out the data and do their own modeling. So you still have access to all of that first party data. You still have access to all the consented users that are kind of opted in, happy for you to track. In a sense, it's become like a CRM on steroids. It's a subset of your data, but these are your high value customers. These are the people that are allowing you to do that because they like you or trust you or a combination of the both. So you can still do your attribution modeling on there, but then also use the other methodologies or techniques like media mix modeling.

So MMM for short, using different sort of analysis and different measurement techniques on the aggregate data. So you can kind of take some learnings on a subset. So do your attribution modeling, do your own kind of models on top of the data that you are able to collect, and then take that learnings and apply that to the, to the whole as well as using different sort of analysis techniques on the, on the other invisible data sets.

Okay. And I suppose, how is, how is, I mean, I don't know if this is in your wheelhouse, but how is advertising changing now? think about you know it was the case that you could you could be a you could be a small business i suppose working with say facebook or google or whatever and you could you could directly market and advertise to to people based on your knowledge of exactly what they do and their exact interests and so on and that means you would get you know you would get marketing materials or messages that were very relevant to you whereas now you look at is it flock which i think the thing from google about about cohorting users based on what i

suppose how is advertising changing in this respect and how are these changes impacting on that and and what experience have you had this really that's a really good question and it's changed a huge amount. And actually it kind of goes back to what you said there about the retargeting. So retargeting is something where you have, you've tracked a single user and you retarget them and you send an advert directly to that user

on another website. So that is the thing that's gone, or at least going. And to be frank, we should think of it as gone already what hasn't gone is contextual targeting so this is where that that kind of flock thing you mentioned from google that was their federated learning of cohorts they've kind of deprecated or parked that idea and they've introduced a new idea which is called topics and they've got a topics api um i'll dig out a link mark so you can stick in the show notes but um what that's basically doing is clustering different people into different um interests and demographics for the for you to then as an advertiser target so rather than

targeting individuals you target similar people with similar interests and that's how you how you run your target your campaigns so in a sense if anyone's aware of i suppose you know even sort of 5 10 15 years ago we'd call that prospecting. You do your prospecting campaigns and retargeting. Prospecting is trying to find new users to your website based on, you know, various different traits like location or gender or other conditions that the advertising platform would have available to you. And retargeting is, here's a person I want to send an advert to, please go do that. So that, like I say, the retargeting is gone. But but in a sense what's happened is that there's been a lot of investment in the prospecting and being able to target groups of users down to um you know all sorts

of different combinations of of of data points so things like facebook and google obviously the big two players here the more data you have the more profiling you can do which means the more measurement that you can do which means you can measure a better roi i suppose ultimately on on their adverts again everything comes back to platforms like google and facebook being able to measure an roi on their ads because they need advertisers to keep spending money with them i mean that's ultimately their their end goal right is to make money with all this stuff okay uh but so maybe to kind of round off this this this conversation really i mean do you feel that so in your opinion is this a good thing or a bad thing or a or you know obviously there's a privacy angle to it which i think people

people at the trending trending society is moving towards that and so on but you think if you're an advertiser if you're a an e-commerce business, do you think the changes that are happening and the new features in GA4 are a good thing or a bad thing? If I had to come down one side of the fence, then I would say a good thing. What ultimately enables them to continue to measure their advertising, right? And again, coming back to Google Analytics being an advertising product,'s what you know the primary focus is for it to do so without this if this wasn't introduced if it was just a kind of the less data the less visibility you had and you kind of have a big blank there basically we kind of we go back into a world where we kind of just run on assumptions and there's nothing to be able to measure and prove anything against and i think

whether or not we trust Google, whether we implicitly trust Google to do the modeling for us, I think the reason I say I come down the side of the fence that this is ultimately a good thing is because, let's be fair, the majority of people, 99 out of 100 organizations, don't have access to data engineers and scientists. So having a platform that can now, in a sense, quote-unquote, do this for free free i think that is ultimately a good thing because you're giving access to you know

good data or good data techniques even if you can't afford the the headcount or the agencies or the support to to be able to do that and what it enables you to do is to continue to invest improve measure you know improve your your business improve your website improve your apps in in lots of different ways the the thing that's kind of holding me back from being you know 100 it's a good thing and you know because i'm not i'm not employed by google right i'm not i'm not even i'm not even a reseller i don't get a cut of anything that i try and sell you and the drawback for me is that it is black box and it is an advertiser that's doing this this is this is a this is an advertising walled garden that has a very, you know, it's got a vested interest

in one thing and that's you spending money through Google. Well, they're marking their own homework, aren't they? Yeah, exactly. I couldn't have said it better myself. Yeah, they're marking their own homework, but they kind of roll this up under this kind of like, oh, it's for your benefit. You know, this is all free stuff for you. And then you think, well, how can Google create a bunch of machine learning models and give,

you know, there's no data limits in GA4 anymore, like we talked about in the previous version. So now you've got an unlimited data collection tool that does loads of machine learning models, and they've given it to you for free. Okay, what's the catch? Of course, there's always a catch, right? And I think that's what's holding me back from being you know 100 enthusiastic about it and i think for me um it's not about there's always there's always an angle there's always a catch every platform you use there's always going to be a limitation or a catch or an angle however for me it's not about you know trying to find the perfect vision of a platform that can give you the perfectly unbiased truth for me it's about just being you know it's an education thing it's just being very aware of what's going on so that you're not blinded by it so that when you get the

kind of the data out the other end you can be you know maybe healthily cynical about the output or the answer and so you can decide to do it's ultimately down to you what you do with the data right google aren't doing anything for you you get to decide what you do with the data and just having a even a very top level view of or an awareness of what each platform again google analytics specifically for this conversation but any platform you use just having a top level view of what they're doing to it can really help you you know not make bad decisions okay and and i suppose one reason once around things off one one reason this is no longer maybe just a theoretical conversation is an announcement

happened a couple of weeks ago that you you mentioned briefly at the start of the uh the conversation about the deprecation of of you've analyzed is about universal analytics so but again just to finish off on that what does that mean and what why is that now why do people now have to start thinking about this really yeah so a couple of weeks ago google announced that they are deprecating or or as they said, sunsetting Universal Analytics as of the 1st of July 2023. What that means is that everyone is being forced, if you want to keep Google Analytics, that is to move over to the new version of Google Analytics 4 by that time. The reason why there's been a bit of a panic right now is because everyone's trying to get Google Analytics 4 set up before the 1st of july 2022 so that come this sort of d-day next year the 1st of

july 2023 you have year-on-year data in the same platform because um as i said at the top actually um there is no historical data you can't transfer one set of data from you know universal analytics to ga4 because it's such a different schema it's such a different data set that you have to just start again and so people are trying to start before the deadline next year so they have that year on year data and what what they've said is as of the 1st of july 2023 you'll um they'll stop processing the data as it comes in so in a sense you'll still have access to the the UI. They said for at least six months. So throughout 2023, you'll still have access to that data via the interface and the API.

But what you won't be able to do is send any more new data into that platform. So in a sense, the 1st of July, 2023 is the kind of final day to have everything done and dusted in terms of like a cultural change, an educational change, making sure everyone's trained up, your advertisers, your marketers, your agencies, all third-party products. I'm sure you're working with a bunch of people pulling that data out

into data warehouses and playing with the data on a larger scale. So there's an awful lot of affected systems from this one announcement, not just Google Analytics, but even taking the data warehouse side, Google Analytics being one data source in, you know, in many, but actually all of the kind of the pipelines, all of the systems, all of the kind of, you know, the transformations will have to have to be adapted and changed over ready for, you know, this time next year, really. Okay. So, so, I mean, I suppose, you know, you said that lifted over and so on. you finding that that your clients are largely just lifting and shifting what they've got or

is this an opportunity to kind of think about what they're doing a bit differently and and and kind of i suppose rework and rethink how they're tracking tracking activity yeah no for sure 100 even if they are thinking of doing a lift and shift i tell them not to so um So, yeah, with this, it's a great opportunity for a fresh start. As we said before, it's like a completely new data schema. It's a different way of thinking about data collection, very similar to the segments and other tools you mentioned, Mark. And actually what I've always found is that the better you've known or you've got to have known universal analytics,

the harder the transition is because universal analytics is a very complicated, very nuanced product um it's very it's very unique and um coming into a world like google analytics 4 that's pretty similar to some other products it's very the scheme is flat uh event structured um it's actually it's actually having to unlearn all of the the workarounds and nuances you learn in universal analytics so in a sense you've got a um a better tracking solution objectively better way of tracking things that happen on your website and app but because we've had the last you know 15 years tracking it in a certain other way it's a it's a transition and i would always say not just because of the data schema but also legacy um you know anything people are always quick to add tracking in,

but they're very, very slow to remove it. So I would always suggest start with a blank sheet of paper and, you know, use the data you already had. What's a KPI? What's important? What does the kind of bare bones look like? Let's move that over, not lift and shift, and almost give ourselves an opportunity to start again. Again, not taking anything away that you need.

It's like decluttering, right? Decluttering your house. It's like, don't get rid of the stuff you actually need and use, but get rid of the stuff you don't use because it's just in the way and a distraction. Okay. So just to round things off then, Dan, it's always been great having you on the show, but just tell us a bit about how people might find out more about about you about measure lab and also the the podcast that you uh you host help host as well yeah for sure um yeah so i mean you know classic website you know

measurelab.co.uk is uh the measure lab website um i do a lot of stuff on there including the podcast so the podcast is the measure pod so it's an analytics focused podcast but quite specifically around google analytics and the google stack and as you can imagine a lot of conversations like So the podcast is the MeasurePod. So it's an analytics-focused podcast, but quite specifically around Google Analytics and the Google Stack. And as you can imagine, a lot of conversations like this around GA4 at the moment. But also my own website. So I have a blog called analytics.co.uk

and I do the occasional post on there with the kind of tips and tricks and helpful guides on some of the concepts around GA4 specifically. Great, great, fantastic. Well, look, it's been really good speaking to you. I've learned a lot myself on this. So that's a really good conversation. Thank you very much for coming on and explaining this kind of world of GA4 and how the industry's changed. And it's been lovely speaking to you. you you you you you you you you you you Terima kasih telah menonton! Thank you.