#131 From raw export to insight-ready data (with Johan van de Werken at GA4Dataform)

[00:00:00] Dara: Hello and welcome back to the Measure Pod on Dara joined as always by Matthew. How are you doing today, Matthew? 

[00:00:21] Matthew: I don’t know. I’m okay. I’m, I’m in one of my, I have these recurring peaks and troughs of existential dread based around various announcements in AI and tinkerings, and I think I’m at one of my peaks currently.

[00:00:35] Dara: It’s all fine. Good. Thank you. I needed that. That’s, it’s fine, EV everything’s fine. Don’t worry about it. Good. It’s fine. I needed that. How are you? I’m good. I’m a bit jealous. I can hear the obvious and easily identifiable sound of a 3D printer behind you, which probably won’t come out in the, in the final, the final cut, final edit.

[00:00:55] Dara: I’m such a professional. I know all these terms, but yeah, there is a, you’ve got a bit of 3D printing going on in the background, haven’t you? 

[00:01:01] Matthew: Yeah, sorry about that. I thought about it earlier today and thought it’d be done and it isn’t and it’s, what did you describe as the sound of a buzz? So in the background it, I have, it did, did 

[00:01:09] Dara: sound, yeah, it did sound a bit like a buzz.

[00:01:10] Matthew: I printed this. It’s a bit of a mask for those listeners, but Yeah, I, I have, I, I started to buck my nerdiness for the 3D printer. 

[00:01:19] Dara: I’m quite enjoying it. Yeah, but listen, fair enough. I said I’m just jealous I don’t have a 3D printer, so. Oh, it stopped. It stopped now. It just stopped. It’s finished. It’s finished.

[00:01:26] Dara: Yeah. Yeah. It heard me. Right. So we’ve got, I think we’ve got a bit of news today, haven’t we? A few, few bits and pieces to go through. Yeah, we, we’ve been I think we can make this little confession. I think our listeners will probably empathize. We were struggling to figure out what happened when, and there’s so much news coming out at the moment and it’s like, hang on.

[00:01:46] Dara: Have we, have we actually not talked about that yet? No, I’m sure we did. no, we haven’t. Yeah, so time is all kind of blurring together. The updates are all blurring together doesn’t help that everything’s called basically the same thing or variations of the same thing. And then people keep changing the names of things as well.

[00:02:01] Dara: So it is pretty difficult to keep up with everything that’s happening in the, especially in the AI space at the moment. 

[00:02:08] Matthew: Yeah. And again, we’ve talked about us just forgetting and killing off companies, but the one we forgot about was open eyes, held an entire dev day. We nearly missed that off our roster because so much other stuff happened that we just forgot about it.

[00:02:24] Dara: Yeah. It was only what, a week ago or something. And we, we were both, we both tuned in live for the keynote. We were chatting about it on Slack. We wrote a bunch of notes and then completely forgot that it even happened. That’s a modern day memory. I rely on AI to do things for me now, not. I, I don’t know if you do this, but I look at the GA4 release notes just to calm myself down every now and again when I get a little overwhelmed.

[00:02:47] Dara: So I just go back to the GA release notes and think, yeah, that’s, that’s a little bit more like it. Yeah. Nice and calm. Feels nice pace. Yeah. So where do we start then? So where you’ve mentioned OpenAI Dev Day, I guess that’s as good a place as any to start. Yeah, I can reel off. 

[00:03:02] Matthew: I don’t think all of them are in-depth discussion worthy, but I can reel off what they talked about.

[00:03:08] Matthew: But essentially a couple of things are added to the API chat. GPT Pro is now in GPT five Pro is now in the API and SAWA two, which we’ll probably talk a bit about in a bit more detail later ’cause that Lovers big fan terrifying and gross in equal measure, is, that’s now in the API. They are releasing loads of new stuff with Codex.

[00:03:29] Matthew: So Codex is like their SDK for devs and building things, and they’ve released Slack integrations and all these kinds of things. But the two main ones, the two sort of ones that were in the keynote, apart, the Codex wasn’t the keynote, but the two main ones that I think d and I globbed onto was agent kit and apps in GPT, catchy name.

[00:03:53] Dara: So where do you wanna start out? Because you said at last, let’s go with the catchy name. And even though it’s not a very catchy name, it’s it, this was probably the most eye-opening one for me anyway. I thought it was, yeah, basically it’s chat, GBT becoming a, a browser, an os everything all rolled into one.

[00:04:12] Dara: Really? 

[00:04:12] Matthew: Yeah. That’s what it feels like. So, so in, in a nutshell, in case people haven’t seen this, they’re essentially making an app ecosystem within chat, GBT. That you can just, the example was like say booking.com, they just typed in, Hey, booking.com, I wanna go here, blah, blah, blah, blah. And it just automatically pulls that app into that context and then it returns and rendered like nice interactive modules is the wrong word, elements within the, within the chat window.

[00:04:40] Matthew: So you could sort of link all these things together. They had stuff around travel. Spotify I think was in there. They’ve got a set of apps which I might be able to quickly find if I am quick. So instead of coming soon, or things like all trails, door Dash, OpenTable, Peloton, TripAdvisor, Uber. So you can imagine just sort of chatting in, in natural language too, I’m going here, I wanna do this, book me an Uber, book me a hotel, and never really leave the chat interface was the, was the big thing. I think. Like, you don’t, you don’t go to the website, you just live there. 

[00:05:18] Dara: Why would you ever need to go anywhere else? Exactly. My first, my first thought on this, it, it, it makes me think back to the philanthropic, I can’t remember what it was called now, but the, like, their browser thing. I’m going to call it a thing.

[00:05:31] Dara: Use a very technical term. The chrome autopilot. The autopilot. That’s the one, yeah. Chrome autopilot. Yeah. And how they were both being quite transparent, which we, we both were kind of fans of seeing how they were willing to share, you know, what wasn’t working and what some of their concerns were.

[00:05:47] Dara: But they were doing it really close, not even in a beta, they just had a very small number of trusted testers. And I do wonder about you, our open air, just like, yeah, we’re just not going to worry about that. Let’s just get people using agents within apps, within agents, whichever way around it is, and not worry too much about it.

[00:06:05] Dara: ’cause this is presumably, this is where there’s going to be an increased risk of prompt injections and, and things like that. 

[00:06:12] Matthew: Well, I don’t know. Well presumably there’s, there’s, there’s some of that. I mean, I think there’s always a risk when you’re putting it to an LLM that it’s going to make something up or, or hallucinate.

[00:06:21] Matthew: I think they’re building apps specifically for, so I think it happens within there rather than going off to the side. But I might be wrong, it’d be interesting. I mean, I think we both said at the time, like it feels like they put all the numbers up on the, on the stage when they were announcing this and it was like 800 million weekly users or something.

[00:06:39] Matthew: And it, and it felt a bit like if you’re a. A website owner or some sort of business that you have to do it. Mm-hmm. You have to go and make an app for chat, GBT because it’s just, it’s like Apple almost when they created their apples, the, the app store for the iPhone in the, in the first instance, and, and then it became the thing.

[00:06:57] Matthew: You had to have an app. Yeah. You gotta be in there on Yeah. On the app store. Yeah. It’s interesting. Yeah. I, I, I’m not, I don’t think it’s available in the UK yet. 

[00:07:06] Dara: No. This is a theme. I tried it. This is a slightly frustrating theme. Yeah. With all of these. It was the same with the instant checkout or whatever.

[00:07:14] Dara: Again, I just can’t, yeah. We talked about it a week ago, but I know the other thing was an agent commerce protocol, but they had another name for instant checkout or something as well. That was the same. It’s the US and Canada. Yeah. So they don’t, which is another 

[00:07:28] Matthew: aspect, another aspect to keeping you in there, isn’t it? You can purchase it there. You can, yeah. You can do everything within reason within there. You get information, ask questions, and do some research. Presumably chain together apps, it, it, you know, trip planner book a hotel, book, a taxi at the other side. 

[00:07:47] Dara: I mean, amazingly powerful and useful, but it makes me think of, do you remember when Google Maps had to start adding messaging because people were basically walking out into traffic and walking off cliffs. Yeah.

[00:07:59] Dara: Drive it into rivers. Yeah. Yeah. And I had to say, you know, please do make sure you actually like, look at where you’re going. Yeah. Like, like the coffee cups, when they say caution, contain hot liquids, you know, it’s like, I think they’re going to have to put something like that in lLMs because nevermind prompt injections, it’s people themselves are probably the bigger risk where they’re just going to be like, yeah, just book me a holiday and it’ll, you know, go and booked at a 50,000 pound luxury holiday.

[00:08:24] Dara: Yeah. And, or you know, they’ll be, yeah, the bank will be ringing ’em up saying, sorry, where’s the, where’s the money you owe us? So, yeah, I think there’s going to have to be some, you know, is that double-edged sword? We talk about this a lot, don’t we? Where. The utility of these things is undeniable, but how much of your own individual responsibility do you wanna hand over to a computer?

[00:08:48] Matthew: Yeah. Yeah. And it’s, I think the answer to that is for most people will be a lot. 

[00:08:52] Dara: Mm, absolutely. Me, me included. Yeah. 

[00:08:57] Matthew: Mean we had the, we’ve had the discussion a little bit with, a little bit of discussion with this, with Brian Clifton a few weeks ago. What it, you know, he, he was more of the opinion that people were pulling back.

[00:09:06] Matthew: And my argument was they are with the current technology, but I, I don’t know if that quite takes into account that this stuff will get better. Yeah. What it knows about you, I don’t think marketing necessarily does, although you get more relevant ads, this stuff will get meaningfully more impactful the more it knows about you. Yeah. And that’s tricky, it is a tricky privacy conversation to have, I suppose. 

[00:09:30] Dara: Definitely. It’s like the privacy aspect has become. Like 3D chess now rather than just regular old fashioned chess. Four D chess maybe, I don’t know. Five d chess. Yeah. We’ll see. We’ll see. Anyway, we’re all going to, I think we’ve set it, we’re going to just hand over everything, all of our information, secrets, history, everything.

[00:09:50] Dara: Just so we don’t have to use different five detailed desktop apps. Yeah. Just, you know, the convenience of being able to do everything in one place outweighs any privacy concerns I could possibly have. 

[00:09:59] Matthew: Yeah. Yeah. I have very little in the first place and now, now I’ve got, now I can order, I can order an Uber by Ty typing it.

[00:10:08] Dara: I could, I could, I could dictate it. I don’t even have to know who types it is. 

[00:10:12] Matthew: Yeah. True Whisper flow. Yeah. Yeah. Not a sponsor of the podcast, but if you listen to whisper flow. 

[00:10:19] Dara: We do need to start asking some of these companies to sponsor us. Although not the ones that we’re trying to take down. They probably won’t wanna sponsor us.

[00:10:25] Matthew: Yeah. But some, the ones who’ve taken down are like JGPT. I think they’d be really scraping the barrel if they were coming to the measure pods for exposure. But maybe I’m wrong. 

[00:10:32] Dara: Maybe, maybe. 

[00:10:34] Matthew: Okay, so what else? What else was that Open OpenAI Dev day? The other big one was the agent kit. So this is like, drag and drop interface that you can chain together, sort of build out little chat agents that you can deploy on your site or whatever.

[00:10:49] Matthew: Just it has a very sort of Zapier or N eight and type interface point, what do you call it, like a graph graphical UI that you can just sort of drag and drop and point to different things if it, if elses. You can point to vector stores, you can do all sorts of stuff and, and it can take people off in different directions, depending on what they’re asking.

[00:11:10] Matthew: But she’s all very cool. Very interesting. The one thing that did strike me, is that Google has had that hidden away in Vertex AI for a while, but it’s just not that accessible or user-friendly. I have found, I think it’s Agent Builder and Agent Kit, like these various tools that exist in there.

[00:11:26] Matthew: But it does feel like OpenAI just. Has been, has done this stuff potentially, ’cause they don’t have the baggage, but just done them in a much more user friendly way than Google. Sometimes maybe different markets are aiming out as well, I suppose, but it was just something that stood out to me.

[00:11:40] Dara: Yeah, you, you said that to me at the time and, I hadn’t, hadn’t known that, which kind of proves your point really. I think, you know, Google, it might be that it’s different. They’re aiming at different markets. You’re right because, you know, it’s not like Google doesn’t have a track record in making very accessible products.

[00:11:55] Dara: Like, just look at ga, you know, they made a completely universally accessible analytics tool that was, you know, that’s on most websites that still exist now for as long as websites exist, which is probably another week or two. So, you know, but for some reason, yeah, they do have some of this functionality in ver ai and, and it’s just not, people don’t know about it, so. Yeah. Yeah. 

[00:12:18] Matthew: I must, I reckon it’s, yeah, they’re going to enterprise or, or medium sized businesses as who they’re servicing and Yeah. And opening eyes might be going for her. Every person who has a tiny business or a up a tiny business and up, 

[00:12:34] Dara: I don’t know. Yeah. Yeah. I think you might be right. Yeah. And then this had the, this is again where the names get confusing, isn’t it?

[00:12:40] Dara: So this is Agent Kit, and then within Agent Kit I think you’ve got chat kit, which is this kind of, you can use a version of chat GPT as like a, as like a module or you know, you can put that inside your own, your own app or agent. Yeah, yeah. And customize it and whatever. 

[00:12:58] Matthew: Yeah. And that’s not to be Confused with any other agent or Kit Kat or Kit No, no. Or, or Kit Kats. Yeah. Or yeah. All the other things the other companies are releasing that all sound exactly the same. Yeah. All slightly different. 

[00:13:10] Dara: Yeah. But it did look cool. Agent Kit, at least in the demo. But you know, it’s always, these demos are like a very, very pre-rehearsed, you know, I think it was like, I’m going to build this in, what did she do in eight minutes or seven minutes or something?

[00:13:24] Matthew: Yeah, but a lot of components were already made and stuff. 

[00:13:27] Dara: Yeah, totally. It was like, yeah. You know, she did a good job of pretending she was nervous about it, but actually it was obviously very well rehearsed. 

[00:13:34] Matthew: Yeah. But, I will give them credit for the lack of clipboard usage compared to Google’s, although Google’s was so heavy.

[00:13:43] Matthew: And then maybe that is the reason, that’s part of the reason why we didn’t see, like the data engineering agent released for another, what, six months from when they announced it because it wasn’t actually ready. But yes, it did just seem to be typing. Maybe they just practiced before. Yeah. To make 

[00:13:59] Dara: sure they got the wording exactly right. I don’t know if, I mean, if I was to give them any advice, I’d say use whisper flow, but you know. Yeah, there we go. Doesn’t look cool though. No, no, no. It doesn’t. That’s true. but you’re right though. They didn’t have the, let me just, quickly do this prompt and then there’s suddenly this block of text lands in.

[00:14:16] Dara: Then you’re like, Hmm. I’m pretty sure you just copy and pasted that. Yeah. So they were the two, they were the two biggies, weren’t they? I think they were probably the ones you did. They are, you did mention Sora too and that, so maybe that’s where we should go next. 

[00:14:27] Matthew: Yeah, I was going to say, do you wanna a pallet cleanser first or do you wanna go straight into Sora too?

[00:14:32] Matthew: Open air? Let’s, let’s just go into, dive into the, into the, into the slop head first. Well, I don’t know where to begin with this one. This, this, this is probably part of me, it’s keeping you open my anxiety. Yeah. Yeah. It’s another video model, first and foremost. Similar to similar, but a lot better than Google’s. I forgot what Google was called now.

[00:14:56] Dara: Exactly. Exactly. Who, who even cares. 

[00:15:00] Matthew: Yeah. VO three so it can generate sound Yeah. And images. And they’re very realistic to the point where a few of them have crossed the uncanny value for me now, and I can’t tell, which is frightening. Yeah. But they’ve wrapped this whole thing up in a TikTok slot, essentially.

[00:15:23] Matthew: So it is a slot, a tick slot, it’s a social media platform essentially, where you can generate videos, you have friends, you can, you can generate videos of you and your friends doing various things. And there’s all sorts of stuff coming out of it that I’ve seen online. Of the one that affected me was there was news people had, someone had stitched together all of these videos they created that was like a news bulletin from the, that could be from anywhere.

[00:15:51] Matthew: It was essentially the zombie apocalypse, but real actual news bulletins and, and. It all just looked so real. I didn’t think a zombie apocalypse was happening, just to be clear. But I did think for the rest of the day, everything I saw, I had this weird feeling in the back of my head like, is this real?

[00:16:11] Matthew: Yeah. And it was a very strange feeling of not trusting anything I saw that was like, That might not be good. 

[00:16:19] Dara: But it is going to be like the boy who cried wolf. Because when the real zombie apocalypse happens, you’re going to just dismiss it and think, oh, here we go. Just more AI slob. 

[00:16:27] Matthew: Yeah. There’s so much moral, and I don’t know what, I don’t know where to begin, but have you, what have you seen of the videos and things that have come out of it? Have you, have you seen much?

[00:16:34] Dara: I, I had a look at that. I mean, I, I’m, I’m really reluctant to, I mean, I, I’ve never consciously gone on TikTok for this very reason, and they are real videos, at least some of them are. I mean, it depends. Yeah, you, you could get into the definition of what, you know, what constitutes real.

[00:16:49] Dara: ’cause I think some of the stuff on TikTok, you could argue Yeah, isn’t, isn’t really real either. But I kind of intentionally stay clear about it, because I just don’t want to get sucked into something like that. So I’m really wary of this as well. So I haven’t gone into the slop beyond just dipping my toe in for safety reasons.

[00:17:07] Dara: But even having not gone in maybe as deep as you did, I’m now doubting things to the point where. You, using your 3D printer, created some awards for a hackathon that the company did last week, and you showed them to me on a video call, and I didn’t believe that they were actually real. I thought you were using some kind of AI video to, to, you know, like a hologram on your hand.

[00:17:32] Dara: And I was thinking they’re no, they’re not. Like you didn’t actually make them. And lo and behold, they were actually real. So they were real. Yeah. So I’m, I’m just doubting things in general now and thinking, you know, if I don’t see it in, in real life even then how can I be? Sure. 

[00:17:46] Matthew: Yeah. I, I, I saw an interesting take of some content creators and there’s, there’s a guy called, Casey Nasin or something along those lines.

[00:17:56] Matthew: He’s a famous vlogger and he was kind of, he kind of laid out this timeline of when you, when people first started creating video too. To now and, and essentially saying the, the walls, the barrier to entry, that the resistance to create has just sort of been evaporating away from, say, film camera to digital and digital to handheld, to handheld, to mobile phones, to platforms like YouTube.

[00:18:22] Matthew: Mm-hmm. YouTube, at least it democratized it enough that people could create things, but there was still a level of sort of creativity, creativity involved, but at the point where all you need to, and maybe TikTok is a level down from that, where it was just very short, algorithmically surfaced, nonsense, nonsense.

[00:18:39] Matthew: And then now when it just needs to be text written, the bio entry has essentially evaporated. Yeah. So the level of crap that’s going to come out of that real realistic looking or not, is going to be pretty unprecedented and could have a. Pretty deep seated and fundamental effect on the Internet’s future and what that looks like down the line.

[00:19:03] Dara: Well, you could, you could almost effectively create a multiverse by saying, I wanna create this video, but do variations based on change the person’s hair color, change what’s in their hand, change this, change that to every different possible combination. And the internet will basically just become a multiverse of slop.

[00:19:21] Matthew: People are already doing it with historical figures and like I sense fairly distasteful alterations like MLK’s speech and some stuff like with Michael Jackson stealing people’s chicken in, in restaurants. Like just bizarre stuff. Yeah. That people were already believing stuff before, like, yeah, ridiculous, outlandish things before this.

[00:19:45] Matthew: Now they’ve got video evidence to back up their, you know, yeah. Prejudice views vice that somebody’s created using. MLK or RFK or whoever else, 

[00:19:56] Dara: It’s true, it is truly terrifying. Yeah. 

[00:19:59] Matthew: This one could be a podcast in of itself I think, but and maybe it will be, but yeah, I mean the good news is that we could probably stop doing this soon and just prompt it. Yeah. We could just podcast. 

[00:20:11] Dara: We Yeah, we could just, yeah. Yeah. I mean, yeah. That’s not a bad idea actually. If you can’t beat ’em, join them. Yeah, exactly. Yeah. Okay. Let’s, let’s, I’ve had enough, we can’t talk about pallet cleansers. Yeah. We need to move away. We need to move away from this. Trying to think of news isn’t creepy. ‘Cause the other one that I was going to go to is this friend.com, but we might, that’s not a palate cleanser, so we might come back to that one. I’ve said it now, right? I think. 

[00:20:33] Matthew: I think, yeah. I think it’s a palate cleanser in that, to vaporware that will ultimately. So what it is, 

[00:20:39] Dara: Is it even real, what you wanna say, what it is? Well, it says the name, you know it like, you wouldn’t pay $1.8 million for the domain name if it, if it wasn’t what it says on the tin. 

[00:20:48] Matthew: Is that where most of the investment went, is the domain? 

[00:20:51] Dara: Friend.com? Yeah. Apparently they raised $2.5 million and they spent 1.8 of that on the domain name friend.com. So it’s like, I don’t know, is it even real, it’s like an, it’s like a necklace that you wear that’s a little AI friend.

[00:21:04] Dara: Yeah. And even, even their own website, which should be marketing it as a useful thing, seems to be suggesting loads of creepy use cases for it. So I, I don’t know. Is it, is it a joke or is it a real, is it real? I think it’s real. 

[00:21:17] Matthew: Yeah. I saw a YouTuber like taking out of a box and interacting with it the other day, but I don’t think it’s like, I’ve already fallen victim to one of these things in the form of the rabbit.

[00:21:29] Dara: R one. The rabbit R one. Yeah. But it’s. 

[00:21:33] Matthew: It is these, it is these attempts at a new form factor with ai, isn’t it? I think the rabbit R one suffered because of it, it could have been an app and probably still can be an app. And most of the big frontier malls now are building stuff like agent kits and apps in GPT are kind of what that was promising and striving to do and computer use.

[00:21:51] Dara: Yeah. 

[00:21:52] Matthew: you had that, the human AI pin that had the little weird projector on it that you shot on your phone, which like the, that’s gone bust Star Trek. Yeah. Yeah. I liked it for the idea that I could shape one like, like a star. Yeah. Star Trek communicator. But that, apart from that, that was it. So yeah, I think it’s just that and it will very quickly go outta business. 

[00:22:14] Dara: Well, I mean, it looks ridiculous. I don’t know if this is just like a pro, the prototype, but why would they have the prototype in their marketing material? But it’s like. It looks like you spend all the marketing budget on the domain.

[00:22:25] Dara: That’s true. They don’t have any money left to make the thing, but it’s like a big pendant that hangs around your neck and they’re like, I, I had to laugh, right? So I just outta curiosity, I just had a little look through the website and there’s an FAQ down the bottom and yeah. Okay. It’s not crazy expensive.

[00:22:42] Dara: It’s $129. I’m not saying that’s not a significant amount of money, but it’s not crazy. First FAQ question, where are my memories stored? All memories are encrypted by your friend’s circuit board. If your friend is lost or damaged, they’re inaccessible forever, right? 

[00:22:56] Matthew: Not great. Question two is, I can’t much, if you get a huge amount of a user’s amount of memories in there either, ’cause it’s like the size of a two big coin or something. 

[00:23:03] Dara: Yeah. And if it’s not backed up or anything, then how, yeah, how, how big could it be? Question two, is my friend waterproof, splash resistant, not swim proof. So, you know, if you’re, if you’re on a particularly rainy day, if you live in the UK or Ireland or anywhere else, don’t, don’t touch, don’t get to your friends.

[00:23:21] Dara: It’s always been my rule. Yeah. I think I’ll keep my $129 and spend it elsewhere. 

[00:23:28] Matthew: Yeah. If there’s going to be, I mean, I was about to say there’s going to be a continuous search for the, for the right form factor. I worry that Meta has figured it out because as regular as I know, I want less, I don’t want meta to win this.

[00:23:42] Matthew: Yeah. Particular AI race, they, I think since the last time we spoke, they’ve released their glasses that actually the 

[00:23:48] Dara: Ray RayBan RayBan glasses.

[00:23:49] Matthew: And interfacing them now like a full color screen. Yeah. That people can, and it’s got AI built into it and stuff. That does feel like a good bet potentially. Maybe not, I don’t know. 

[00:23:59] Dara: But yeah, there’s the Johnny Ice Sam Alman collab. Yeah. Mystery. Yeah. I’m still waiting to hear if anything happens with that. 

[00:24:09] Matthew: Just a triangular phone or something Probably. 

[00:24:12] Dara: Yeah. And everyone will buy it for 2000 pounds or whatever it’s going to be. It’s a pi, maybe it could be a pyramid. You just carry around with you. A small pyramid. Two, two or three feet tall, you know, not that big. Yeah. Yeah. But it’s got all the models in Bill on board. Yeah, yeah, yeah, yeah. We’ll see, we’ll see what happens to that one. But, I’m going to give friend.com a miss, and this episode was sponsored by friend.com.

[00:24:37] Dara: That’s a no. What else did we get? So, oh, we can confirm. We can just desalt a piece of news. So we shared the rumor that five Chan might be buying DBT and it turns out they’re merging. Yes. Which I think means they’re buying. 

[00:24:56] Matthew: Strange, isn’t it? Yeah. The language very much is merging and so on, but they, and we talk about this today, by the way, in today’s podcast, so we will be referencing the fact that there might be a merge, but it has happened when we’re recording the introduction.

[00:25:09] Matthew: But yeah, they are merging into something like a 600 value company with a combined RR of 600 billion or something million I think. 

[00:25:18] Dara: Yeah. Annual recurring revenue of 600 million. 

[00:25:21] Matthew: Yeah, million, sorry, not billion. Yeah. Yeah. 

[00:25:23] Dara: But it’s a, that’s, yeah. Substantial, isn’t it? 

[00:25:26] Matthew: With a customer base of well over 10,000 customers that feels low. Were those companies? 

[00:25:30] Dara: Yeah. I don’t know. I don’t know. 

[00:25:32] Matthew: Yeah. Apparently the apps as soon as remain the same right now, and I imagine this is what a lot of companies say when they merge. There’s probably a very set thing you have to say, but. Five Tran DBT communities will remain completely separate and will still be striving towards their roadmaps or whatever they’re wanting to build.

[00:25:52] Matthew: Yeah. But the real excitement is what we can build together. That’s what they say in their press release. I mean, you’ve got me excited. Yeah. I, I don’t believe in much of what our lead will read in a merger press release. Like they’ve gotta, they’ve gotta try and put out fires and worries about Yeah.

[00:26:07] Matthew: Exactly. What the merger means and things. I think there’s still a, still an, an, an interesting thought about that independence of DBT kinda going away. 

[00:26:15] Dara: Yeah. I think that’s, yeah. Be interesting to see whether that actually has an impact or, or, or not. ’cause you know, obviously the, the kind of vibrant DBT community and it was the original in the space.

[00:26:27] Dara: I don’t know, maybe things have changed since then. Maybe people won’t care as much about that. These things happen, don’t they? Companies merge. They get bought, but we’ll, we’ll see. And then as a final, well, I think final piece of news, maybe I’ve forgotten something, like we almost forgot Dev Day, but Google, in true Google fashion have decided to change the name of something. So Agent Space is now Gemini Enterprise? Yes. Or is it part of Gemini Enterprise? 

[00:26:54] Matthew: Yeah. I can’t quite tell. It is in true Google fashion. 

[00:26:58] Dara: I asked Gemini. 

[00:26:59] Matthew: Oh, did you? 

[00:27:00] Dara: Yeah. 

[00:27:00] Matthew: What did Gemini say? 

[00:27:01] Dara: I’ll tell you. I don’t know when I opened up the tab. Yes. Agent Space is rebranded to Gemini Enterprise. This is more than just a name change.

[00:27:08] Dara: It represents a significant expansion of Google’s AI powered enterprise offerings. And then it goes on to give breakdown, unified platform made by Google agents, no code workbench, deeper workspace integration, enhanced multimodality, centralized governance and marketplace and so on. But can we believe this? ‘Cause it’s from ai, so maybe it’s. 

[00:27:28] Matthew: Yeah. I, I, I’m, I read the marketing spiel, I kind of got a feeling it was, there was more workspace stuff in there. Yeah. The annoying thing. A really annoying thing. And it took going back to our AI legal seven, 

[00:27:42] Dara: seven months ago or something. 

[00:27:43] Matthew: Yeah, exactly. But so open air, all the stuff we mentioned, can’t access it.

[00:27:48] Matthew: Google Enterprise released and talked about however long this is on, off the back of agent space that we still haven’t really Yeah. Managed to get in. I went to sign up for it yesterday. We’ve, we’ll let you know. Yeah. When it’s, when the space is available. Don’t, don’t call 

[00:28:02] Dara: us. We’ll call you. 

[00:28:03] Matthew: Yeah. For God’s sake. Sometimes I don’t wanna live in America, but. 

[00:28:07] Dara: It’d be nice just to, it might be worth it. Yeah. Over there announcements. Yeah. Yeah. Just sign up for things and, yeah. 

[00:28:13] Matthew: Yeah. It is frustrating, but it does look interesting. I can see it being, just being able to join all those things up. I think you can build agents there.

[00:28:20] Matthew: You’ve obviously got easy local access to various components, knowledge, and a centralized knowledge repository for organizations, which could be really useful from a business perspective or from a coding perspective. 

[00:28:34] Dara: So it does look, yeah, good. 

[00:28:37] Matthew: If you add, could advocate in it and have a look at it properly.

[00:28:39] Dara: And I guess it does bulk out workspace, doesn’t it? It’s like they really are gunning for these enterprise accounts, which, you know, they, they like to be able to bolster on all the AI capabilities and the power of the cloud behind it as well. It’s, you know, they’re often game, it seems so obvious.

[00:28:56] Matthew: Really. Yeah. That you, that it would, you know, if you are in a dark and you’re typing away at something, essentially it’s the ultimate form of clippy from old Microsoft Word, where I could just say. You’re writing a scope about this. You’ve also written these five scopes previously. Would you like to blah, blah, blah and pull some of that in?

[00:29:13] Matthew: Or you’re doing this and you’ve got, you’ve, you’ve received this email. Just pulling all that information together and making it more useful. It seems like an obvious sort of dream, and maybe that’s what this is. I don’t know. 

[00:29:23] Dara: I got you. I, I have to admit, I got slightly distracted there thinking about clippy.

[00:29:28] Matthew: I thought you looked a bit, you wandered up in your head. 

[00:29:31] Dara: I, I did. Yeah. Yeah. Oh, clippy. What a guy. 

[00:29:34] Matthew: Wasn’t useless. Annoying assistant. That is so useless. What were they trying to do? Maybe they just wanted to pretend they had AI back in 1994, but they didn’t. They were trailblazers. Yeah. We just didn’t understand what Clippy was offering.

[00:29:47] Matthew: Don’t mind copilot Microsoft. Bring back clip clippy as your system. Bring back clippy. Yeah. The one other thing I saw, and so I’ll do this very quickly ’cause I got a lot of news this week. Yeah. Salty in for a while. Gemini 3.0 yes. Seems like it’s imminent or whether they’re going to call it Gemini 3.0.

[00:30:04] Matthew: I don’t know. But another Google model is doing the rounds in various early testing and comparison playgrounds and things. We say this every time. It looks very powerful from what people are seeing. For example, someone tried to get a GPT five, sonnet 4.5, and Gemini 3.0. I did that in air quotes when anyone was listening to just make windows like a one shot attempt at creating a Windows operating system.

[00:30:32] Matthew: I believe that’s what it was doing. Anyway, it was a kind of truncated video and Gemini 3.0. It was pretty much windows that it spat out at the other end, which was pretty impressive. The other two were not windows. Not Windows looking, not windows. Things. Take all that with a pinch of salt. Just some random thing I saw on a LinkedIn feed, but this is something new, so that’s fine.

[00:30:51] Matthew: But yeah, the predictions have always been that Google would be leading the. Model race by the end of this year. 

[00:30:57] Dara: Yeah. Mark Edmondson to be believed, didn’t he? 

[00:30:59] Matthew: Yeah. 

[00:31:00] Dara: Yeah. 

[00:31:00] Matthew: Yeah. So there’s rumors. October 22nd I’ll chuck it out there as a, ’cause I, if it is October 22nd, I might actually get in ahead of that prediction this time because I’ve done three other predictions that have all been kind of true. But the podcast came out after they happened, so just maybe it looked like it was Yeah. 

[00:31:18] Dara: And they weren’t and they weren’t true until you went back and rerecorded Yeah. Edited it. Yeah. Listen, I’m not that impressed with the Windows thing ’cause they made Super Mario Brothers in Excel years ago, so, I mean, whatever.

[00:31:29] Dara: That’s true. Yeah. Yeah. Don’t bother. Yeah. Is that what they call progress? What about Super Mario? 64 in Excel. Come on guys. Oh, now exactly. There you go. Okay. That’s a wrap for a kind of quite busy news. Newsweek really. So our guest today is a second timer on the Measure pod. Johan Vander Kin, when he was on before he talked to us about his kind of learning platform, I guess you’d call it, which was GA for BigQuery, which had a lot of resources around kinda SQL scripts and various bits and pieces for working with the GA4 data in BigQuery.

[00:32:05] Dara: And his focus now is on GA4 data form. So, it was good to have him back on. It was quite a while since he was on the time before, but it was a good, good chat catching up with what he’s working on now and getting some of his thoughts on our favorite themes. So you know, how AI is going to disrupt or not disrupt everything or nothing.

[00:32:26] Matthew: Yeah. Particularly satisfying answers to the next two years as well I, I found. 

[00:32:31] Dara: Different answer. Yeah. Different answers. Yeah. Yeah. Yeah. Very interesting. A nice little setup there. You’re going to make people listen to the whole episode, Matt, to find out what he says. I’m really getting angry about this. Enjoy.

[00:32:41] Dara: Enjoy joining us on the Measure Pod today. For the second time we have Johan Vander Kin. So I had to look this up, but Johan, when you were on previously, it was episode number 72, and that was back in March 23. So it was two and a half years ago. But we’re really excited to have you back on. So welcome back to the measure part.

[00:33:02] Johan: Thanks. What episode is this? Like 100 something. 

[00:33:06] Dara: Oh, now, now you’ve, I pause on the spot there. Yeah. Now you, now, now you’ve asked a difficult question. This will be the 31st. Wow. So you’re around about the halfway point, give or take. And obviously the world has changed dramatically in that time. 

[00:33:21] Johan: It did. So we’re talking 2023 when I was last on then, right? Yeah. 

[00:33:26] Dara: Yeah. Wow. Yeah. March 23. Yeah. 

[00:33:29] Johan: Yeah. 

[00:33:30] Dara: So that time we talked a lot about GA for BigQuery, but actually, you’ve got a new project these days, which is GA for data form. Yep. So, before we get into that, maybe it’s worth just, I know a lot of our listeners have obviously listened from the very beginning and they will have heard your intro before, but maybe there’s some newer listeners who don’t know who you are, Johan.

[00:33:50] Dara: So would you mind giving just a bit of an intro to yourself, a little bit about your background leading us up to what you’re doing today? 

[00:33:58] Johan: Yeah, no problem. So I was just thinking while you asked the question, I’m not sure I know who I am, but there’s another, that’s for another, for another moment.

[00:34:10] Johan: So, yeah, I, I’m Yolo, I’m 43 at the moment. I live in the Netherlands, close to, I’ve been in the digital analytics community for about 15 years now. I guess I used to be a journalist before that, but that’s a long, long time ago when data journalism didn’t exist. Otherwise, I would probably end up there still thinking about it once in a while to do that.

[00:34:37] Johan: Yeah, so I started at a digital marketing agency. It just, for me, was a career change. So I was open to basically anything involving the skills that I had in my backpack. So at that time I could pick between web analytics and technical. SEO could have gone either way. I ended up in a web analytics team.

[00:34:58] Johan: You made the right choice. I don’t know. SEO is dead. Right. So I hear you again. Again, again, again. 

[00:35:05] Matthew: Yeah. That version of it is anyway. 

[00:35:08] Johan: Yeah. So I think Jim, the show just posted an article, digital Analytics, you’ve dad, so I don’t know, what’s a better choice? Anyway, so I started really at, you know, tagging Google Analytics, Adobe Analytics, dashboarding back then mainly in Data Studio.

[00:35:29] Johan: A lot of it was powered by spreadsheets, Google Sheets. So I started learning every possible Google Sheets function. That was to automate stuff basically. And eventually the data sets got bigger and bigger. We had some enterprise clients, so those Google, especially Google Analytics, 360 clients had access to the BigQuery exports.

[00:35:55] Johan: it was still in the days of universal analytics, of course. And so that was my entry to BigQuery and in general, data warehouses and, and everything, you know, data transformation was a completely new world for me. So first I treated it as a big spreadsheet, a bigger spreadsheet. From there, I learned really a lot and basically, I was focused most on building dashboards and automating stuff for our clients.

[00:36:23] Johan: Then I became very interested in that Google Analytics export. I searched for information, I learned SQL, but I was so little content on that, so I started to write about it. Eventually that led to my own platform, g4 big query.com. First, it only contained Juno Universal Analytics, but then I, when the analytics web app or Apple Web, which later became G four came out, I grabbed that domain name, which was not a bad choice, in hindsight.

[00:36:55] Johan: And that platform became like the go-to place for any information about the Google Analytics exports. And kids know this was before LLMs, right? There was a lot of demand. Mm. Nowadays I think you can use practical, practically any LLM to, to get you started and still it’s questionable what your results will be, you know, so we decided, and, and lately I’ve been partnering with, Ash from Hungary is from, marketing Lens, which is an, an, agency in Hungary.

[00:37:35] Johan: And we partnered up to keep the platform going. Then, we are broadening the scope more to BigQuery data form as well as including the Google Ads data and that kind of stuff to make it more relevant, both to keep you updated about everything that’s going on, because it’s a lot. And we also have a newsletter for you, so you can sign up for free.

[00:37:58] Johan: You’ll get a monthly newsletter with me, I think. Almost every valuable article that has come out that month. And that’s also the premium section where you will find more in-depth tutorials and that kind of, so the platform still lives, it still has an audience. I still think it’s a good addition to every resource you can find for free nowadays, especially because, you know, LLMs need to be trained on something and it cannot figure out stuff by itself, at least not in this particular niche, I guess.

[00:38:36] Johan: And the goal analytics exports and, and the goal Edge dashboards are very, very complex. You know, there’s a lot of in and outs you have to know to deal with it. And then, also when it comes to the automation part, there’s a new thing on the horizon and that’s data form. And, and that’s probably what. Brings me here, I guess. 

[00:38:57] Dara: I think it is. But before we, I, I’m just really curious ’cause you mentioned that the audience at least was starting to decline on GA for BigQuery because of the introduction of LLMs. But just at the end there you kind of suggested at least that there is still something that that platform is offering that maybe you can’t get from LLMs.

[00:39:17] Dara: Do you think that is the case? If somebody had the right initiative, could they just use chat or Claude or whatever to get what they could get from J for BigQuery? Or do you think there is still something. Platform. And I’m actually, I’m probably kind of thinking more broadly, even gen even than just be quick.

[00:39:34] Dara: Sure. And maybe just training material, and knowledge in general. Do you think there is still a place for learning platforms? 

[00:39:42] Johan: Yeah, definitely. So that’s, that’s maybe a good bridge to, another project of mine, which is the course I built for the second time for Team Zimmer, which is a learning platform by Smar and Mari Aha.

[00:39:55] Johan: well known in the industry. And that shows there is still demand. We just released a completely new series of courses, like very large courses or hundreds of attempted videos. I guess hours and hours of material. It’s all manually recorded. No AI involved, you know, so very old fashioned, I would say purely based on my experience and their CMOs experience.

[00:40:25] Johan: What I do there is open the screen. You see me clicking in the interface, building queries, explaining from the very beginning of the first lines of SQL until very complex, you know, attribution, attribution stuff, a more complex queries and a new course that we added, which is really different from the first time we did that is of course about, truly about integrations and automation.

[00:40:51] Johan: And that is really a topic that is still very, I would say it’s very, the entry level is quite high, from most people. From a marketing background. I’m not sure they could do it on their own, just by using lms. And the demand we see for the course now is to prove that we are right. ’cause we were, frankly, we were a bit.

[00:41:14] Johan: Trade is a big word, but you know, is there still demand? And, and I think we for now, we proved that there’s still demand for, for high qualitative courses and learning materials. 

[00:41:25] Matthew: I suppose like people who don’t describe the sort of length of the journey you take people on in that course from first click to complex attribution, a novice might.

[00:41:36] Matthew: It’s all about that domain knowledge thing, isn’t it? That, that we keep saying is, is going to be the last, the last island in all of this sort of stuff that’s getting sold by lms. People don’t know the questions to ask in the first place. You can go, you can have a box that can just give you any answer you could possibly want, but if you don’t know the right things to ask to get you the skills you need, then that’s tho those curated domain specific pieces like the website and like the course. Surely for at least for now, that feels like there’s still a place you would hope. 

[00:42:05] Johan: Definitely, yeah. That’s what we’re, what we’re seeing. Yeah. So at the same time, and that. Yeah, I don’t know. Probably you, you were going to ask for it, but I, I was thinking a lot about, you know, automation and then data form came around the corner.

[00:42:22] Johan: So basically a way to, to automate different scheduled queries. That’s how you can summarize it. We don’t have to worry about the order of queries to, to go from completely raw data to, to a data set that is like aggregated at the, at the right level for your dashboards. It has data quality checks and everything.

[00:42:44] Johan: So that tool came around the corner integrated in Google Cloud projects. And then I was like, okay, am I going to, to automate the stuff I write about in, on tier four BigQuery and am I canalizing on my own? You know, I just, what do, what do I need to do with this? And, and that was interesting at the time, I guess I figured that out in the last couple of years.

[00:43:10] Matthew: What was your journey then with Data Form? Did you just see it appear and immediately you just saw its be, I mean, I know you mentioned very early on that you were always looking at automations and augmentations. Is it, did you see the benefit of it straight away and just delve straight in? Or was it a bit of a slower process?

[00:43:28] Johan: So what really helped me is that, you know, in the meantime, I was still employed, so my last employment, so I, I’ve been a freelancer for two and a half years now. But before that, I, I’m not quite sure the last time we spoke, maybe I was still employed or maybe just became a freelancer at that point.

[00:43:50] Johan: But it wasn’t 2023. I was working at a large Dutch bank. They were not, they had Google Analytics and, you know, Google ads and that kind of stuff. But everything else was in AWS. So I was not in the BigQuery or Google Cloud environment at all. It was AWS and so for data transformation we used DBT, and that’s basically data formed from gender agnostic, I would say for now.

[00:44:21] Johan: For, oh yeah, that’s for now. There are some rumors that Pfizer and, yeah, wants to acquire them, but still we use DPT Core, which is like an open source thing. And, and that’s basically the, that was a revolutionary tool and it was not only a tool, it was also, I would say a movement almost. It was a way of working and that really opened up a lot of possibilities for me.

[00:44:51] Johan: And it’s, I learned so much about, you know, data transformation, in a broader context. ’cause I, I’ve always been in the data transformation. I’ve always worked in that field, I would say, because you can, what you do in a goal sheet. Using formulas is also data transformation. In my, yeah, in my shoe. But this was more in a structured way, you know, also with data quality checks, documentation, everything.

[00:45:17] Johan: So when I, and I know Data Form already existed, it was just a competitor of GBT at the time. But then Google eventually acquired them, integrated it in Google Cloud, and then I was like, oh 1 0 1 is three. You know, that moment of realization. So of course I started experimenting with it, right away at first thought, okay, I can just create a bunch of tutorials for G four big grade.com and that’s it.

[00:45:46] Johan: Right? And then eventually I went to, for the first time, to super week in a digital analytic analytics conference also in Hungary. Matt was there, not that year, I guess. No. 

[00:46:01] Matthew: I think the year before, I think it was only. 

[00:46:03] Johan: So this was early 2024. Then I met a couple of other guys there, some I already knew. And we were at, presentation of a Google employee that was presenting an open source rep day built on getup. I think it was called Q4 data form as well, something like that, Q4 dash data form, something like that.

[00:46:28] Johan: It still exists and was basically a way to deal with Q4 BigQuery exports, using data form, but it was very basic. And, but you know, it was the, the same idea was already working in my hat. And when we attended that session, we were like, okay, this is one of those Google unofficial Google experiments.

[00:46:53] Johan: They release it and they will never look back. And that’s exactly what happened. So we looked each other in the eye and I was like, Hey, we can do this. We can do it better. And, you know, make it the standard for the industry because Google will never, probably never, never say never. But I, I think they will, they will not go all in on this.

[00:47:17] Johan: So, because they’re moving on to the next probably AI related thing. 

[00:47:22] Matthew: Yeah, I was probably on the cusp of them changing everything to be LLM focused. Yeah, pretty much. 

[00:47:28] Johan: Yeah. So basically what that solution did was flattening the event table to a session table, more or less, some attribution. It was quite basic, but it was a good starting point.

[00:47:42] Johan: So we got inspired and the guys wanted to, we teamed up with four other guys. We wanted to release it as a, you know, community project. And that’s what we did. We made a product. But the main difference, or the main addition basically, is that we also created an installer. So it’s user interface. You log in with your Google accounts, you do a couple of clicks and your whole data form repository, repository is set up.

[00:48:11] Johan: The code is there. If you want, we can already run it for the first time workflow and boom, you have a ready to use Google Analytics session table, platinum Event table, and also Transaction table. There you go. Three. And that’s what we have now. We released it about a year ago, right? I think we have the M1 L.

[00:48:33] Johan: Of course, we released it using the domain name, G four data form.com that I already claimed. Because I had some experience in how important domain names like that can be. I am the, the usage is quite good. So we have a lot of users, also big companies using our core version as we call it. So it’s a freemium model.

[00:48:56] Johan: we have a core version that will always be free. It’s very generous because it does not only flatten your tables, but also handles incrementally, sorry. And that’s another big advantage of using data form is that you don’t have to recalculate the historical data over and over again, like you probably would do with scheduled queries.

[00:49:21] Johan: When you ask lms, they’ll make you do that if they don’t care about, you know, optimizing for cost and performance. We do so, so our goal is really to set industry standards for these very niches. Dataset that everyone that uses G four has access to for free. So, we have some quite idealistic guys in our team, so we also wanted to make it really open source as well.

[00:49:49] Johan: We call that G four Data Forum community. It is arapo on GitHub and you can use it, adapt it, whatever. And, for a couple of weeks we have released G four Terraform Premium, which is, you know, a premium solution that will bring extra features, support like expert support. We are basically five guys and, together with some others, maybe there are five or 10 others on the planet that have the same knowledge as we have.

[00:50:25] Johan: It’s very niche, but I think there’s value in this. So that’s why we also decided to start a real business around it. 

[00:50:34] Dara: You mentioned there, Yohana, you’ve got some big companies using, using the core, the core version. I’m curious just what the, what the kind of typical customer is.

[00:50:44] Johan: So have you, that’s a good question. We are, we are, we are figuring that out at the moment. That’s always, we’re still at the early phase. We are shared in a very early phase. So of course the core version and especially the community version, we don’t even know who uses that, right? Yeah, yeah. We don’t know. And that can be a really big organization because typically they have their own data engineers just using arapo on and adapting and you’ll never hear from them again.

[00:51:14] Johan: Right. So we don’t know. But we, we think it’s, it’s the good thing to do, to just provide it to the community. 

[00:51:20] Dara: Yeah. Do you think those companies, I know again, you, Kate, if you don’t know who they are, you don’t know this, but do you think that they’re using it as a kind of foundational layer and then that they’re building on top of that? So it’s almost like giving them a jumpstart in a way? 

[00:51:33] Johan: That’s the exact idea. Yeah. So there’s, so for DBT, there has been a project around for a long time that’s doing the same as we do, but it could, you know, it’s, it’s only code, it’s only the data models. So I would say for the digital analytics scene, especially organizations that don’t have a dedicated data engineer, our solution with Core, which is not technically open source, but it’s, you know, you only have to give your email address to us.

[00:52:03] Johan: And that’s, that’s the only thing. But then you can use the installer and we do all the, you know, the hard stuff, the configuration stuff, etc. That is really, you know, unique to what we offer, and that’s for free. Right? So with the basic core version, you already have a great starting point to start analyzing, start connecting dashboards even to the, to the dataset.

[00:52:31] Johan: So no one needs to use premium. I know it’s a bad sales thing to say, but I really believe our core version is so good that you don’t need a premium. 

[00:52:42] Matthew: I suppose this might have been why premium came to be in the first place. I don’t know, but. My question was going to be, can you abstract away the complexity of the setup and the flattening and the getting the data in there.

[00:52:53] Matthew: But obviously a lot of the people who are going to be using this tool might just be marketing tools. And to your own point, you’re like, don’t have a, sorry, marketing team. And to you, they might not have a data engineer. I wonder what their reaction is post all of that happening and then going into data form and looking at it.

[00:53:08] Matthew: Are they like, it works, but what the hell is this? Do you know what I mean? Yeah. Isn’t that part of the idea around premiums? You’ve got that support to help you figure out what you’ve now got, what power you’ve now got You. Indeed. 

[00:53:20] Johan: So we, we give you an onboarding call, we can actually configure it for you. It’s all included in a premium, in a premium plan. And we can even advise, consult and, but also build custom models on top of your standard code as we say. So we also, in the code, the way the code is structured, we use folders. And we have core and custom. So everything you place in custom, we’ll never touch that because the crazy thing about GF four Data Farm is if we, if we update the products, you can go to the installer again, it’ll detect which version you have installed.

[00:54:01] Johan: You’ll say, Hey, this one is old. I have a new one for you. Let me update it for you. And then we update the core piles, but never the custom piles so you can safely, you know, use the installer and we handle all that stuff for you. So that, and we, we update to Pigs Box, obviously, but also to add new features.

[00:54:22] Johan: We still do that with the core version as well. So eventually the idea is we release premium features and after, you know, after an amount of time, some of those premium features will become standard. Maybe even can be incorporated in the core version as well. But, the big differentiator is for premium if you have access to us to support.

[00:54:44] Dara: How far does that support go? So if you, if you have somebody who say, has no, no skills in this area, but their marketing person, they know what data they want, they know what they want have in the dashboard and they know that there’s power in getting that export from GA4 into BigQuery and then doing something with it.

[00:55:01] Dara: Would you create those end products for them? Would you create dashboards for them to use? No. So it’s just the models and it’s getting it. 

[00:55:09] Johan: Yeah. So we thought about that. if, if we would do that, we would become an agency. Yeah. Right. True. And we don’t want to become an agency. We worked at agencies.

[00:55:20] Johan: Some of us still work at agencies like freelancers. There’s a reason we don’t want to become an agency. No.

[00:55:30] Johan: There are, they’re like, you know, we are all freelancers. We enjoy that. Yeah, we can do custom stuff, but custom stuff will be, and we also offer a custom plan, which is basically premium with, you know, we can, negotiate, a track based on a, a scope or a project. We will do that for a prize. You know, we can, that’s, that’s how an A GT also works.

[00:55:54] Johan: But we will only build models and offer, like, can build custom models. So the data transformation stuff, but we’re not going to create a dashboard for you that’s for others to take care of. And what we also do is, you know, offer premium advice so we can, and so not premium, but custom advice.

[00:56:18] Johan: So we can, we will take that very broadly. If you are interested in starting a data warehouse and you want to connect other marketing sources to it. Then we can advise around that. But, and, and you know, for the first, premium clients, we are at the moment looking into, building, standard models for Google ads transfer, that kind of stuff. So do you obviously know the low hanging fruit, so to say. 

[00:56:45] Dara: Yeah. ’cause I guess where my, where my, that question came from and, and you’ve kind of answered it now, I’m thinking of those people who maybe know GA4 pretty well. They know Looker Studio, but the Gap, the gap is in between the two. They don’t really know what to do with those nested tables.

[00:57:01] Dara: So if you are somebody like that who’s sitting on your own in an organization, you are the analytics person, you’ve gotta figure all this stuff out, this could really help bridge that gap. So I know, I know what I want at the end of it and I know what’s going into it. I just don’t have the skills for that bit in the middle.

[00:57:16] Dara: They could use the core version, it wouldn’t cost them any money. They get the nice clean business ready tables that they can then set up with their own Looker Studio dashboards. And it just takes that bit in the middle, that kind of more niche skillset, that specialist knowledge. So for those kinds of marketers, it would be ideal for them.

[00:57:37] Johan: Yeah. So, and if that is not, there’s a possibility that’s still too hard to do because with the core version will, you’ll get very clean base tables and some, you know, session, aggregated session tables, but probably just not yet what you’re looking for or it’s not aggregated enough or, you know, there’s always custom questions from business owners that like the average marketer is not capable of answering with our base tables, you will get with core.

[00:58:09] Johan: So in that case, and we’re still thinking about that. We’re really at the early phase of our company at the moment. We’re thinking about, you know, how we can provide a place for partners, like build an ecosystem of partners around us that can handle those specific questions between the end product and what, you know, what we deliver.

[00:58:34] Johan: There’s a gap. I agree with that. What, we are not here to fill that gap. 

[00:58:38] Matthew: I mean, interestingly, you know, we’ve probably worked with three, four clients that have come to us with core GA4 data form, and we’ve built custom aspects on top of it for them to ultimately build them dashboards and things like that.

[00:58:51] Matthew: So it, you know, it that kind of, oh, it’s kind of, I guess, happening, behind the scenes where the people have probably, it’s just, it’s, it’s just it and not official yet then I, 

[00:59:00] Johan: I guess Yes. Yes. 

[00:59:02] Matthew: Yeah. Gimme a hook. Gimme a ring for it becomes official. Please. 

[00:59:06] Johan: I will get you, you know, a batch partner badge.

[00:59:08] Matthew: Yes. Thank you. Yeah. Good pay for it, but, you know. Yeah. But yeah, it’s interesting. It is interesting that people are, like you say, they’re going, they’re grabbing it, they’re seeing the, the power and utility of it, but they’re just, they’re just not knowing how to take that next step to potentially add those custom bits or build the dashboards or know how to join whatever together.

[00:59:29] Johan: I really, really, I really believe this is, you know, creating a, like a new picture for everyone. So also agencies can, if they are smart, they, like you do, they jump on this topic and they need not, build everything on their own like we did, but use us as a, as a basis or, you know, build on top of that because there’s still a gap and that gap is not going to be served by us.

[01:00:03] Matthew: I wonder if it will be served by ai, which is my segue into something. That I’ve noticed that I’ve ju I’ve been doing data form projects recently and what has appeared within Data Form itself, which I didn’t quite appreciate was going to appear, is that the data engineering agent, I don’t know if you’ve, you’ve seen this and I, like, I knew it was available in the, in the pipeline section, you could get it to sort of do pipelines and I knew that was built on top of data form.

[01:00:34] Matthew: What I didn’t quite appreciate is the inside of the data form, it was editing my data form files and, and fiddling around with stuff. Have you seen that and played with that and what’s your kind of experience with it? 

[01:00:48] Johan: I’ve not seen it myself yet. ’cause at G four Data form, I am focusing more on the marketing side now, but the other guys probably have, but in general, I think we’re at a stage where companies are just, you know, integrating AI everything and there will be a point that they will realize and, and use this.

[01:01:10] Johan: Also realize that a lot in some parts of you don’t want to have it because especially a data form, you don’t want an agent, you know, an autonomous agent, changing configuration stuff without you knowing it. That’s not something I would want. I want to have control there. And yes, if I have a question, it’s nice to have a place where I can ask it, but I don’t trust the quality of it yet.

[01:01:40] Johan: And especially Google’s, you know, they have proved over and over again that just try the Gemini sheets. You want to jump out of a window. She was there. It’s not for a simple sum of, of a one column or so, but it doesn’t make sense yet. It’s too early. Mm. Maybe there will be a point when it’s good enough, but it’s not now.

[01:02:04] Johan: So I agree. I don’t like it at this. 

[01:02:07] Dara: What about a, a ga per data form MCP server? Have you thought about that? Or maybe even somebody’s built one? 

[01:02:15] Johan: Well, we’re focusing on the, yeah, on the basics first before we, so what we’re more, what we’re more looking at, first ’cause you ask who, who is your audience, basically?

[01:02:30] Johan: And one of those is, obviously , agencies as well. Right? So, so we, we have brands that can have multiple tier four properties running. So we still need, we are in the process of building multiple properties support, but we also want to, we are looking into building a user interface for agencies to manage multiple installations of G four data form, that kind of stuff.

[01:02:55] Johan: So to do stuff in bulk, that’s, that’s things we are looking into now rather than, what you are mm-hmm. Suggesting 

[01:03:05] Matthew: I built. Not for GA4 data form, but I’ve done a data form one data form documentation MCP. Cool. You can have, you can have that Johan if you want. You heard you heard it here first? 

[01:03:15] Dara: Yes. How do you balance the, I, I was kind of interested to, ’cause I, you even said this very honestly earlier, you said you think everybody can just use the core version.

[01:03:26] Dara: So how, how do you balance, ’cause obviously you do have the premium features and you presumably would want to get rich. You do. Exactly. Yeah. Yeah. You do want to get rich. So how are you? Yes. How, how are you finding that balance? So you, do you find that easy and even amongst the five of you, do you, do you feel like there’s a little bit, you know, is it all quite balanced in terms of how much of it you push towards the community versus how much of it is going to make you all rich and retire early?

[01:03:51] Johan: Yeah, we retire early. We’re not young, so we have to do it quickly. So many, I think we have a very unique team in the sense that we are all nerds. We’ve met at super week with, it’s basically one big nerd chest. I was the one that came with the idea, Hey, there’s value in this. Let’s eventually ask for money somewhere down the road.

[01:04:18] Johan: The other guys just wanted to create an episode thing. So I’m the commercial guy here. That says a lot about our company. I guess also we have a motto and that is, don’t be a dick. That’s basically our strategy. Yes, we want to get rich. Yes, we want to, you know, we really believe there’s value, but not at the cost of everything.

[01:04:43] Johan: We really believe we are. Everything we do on a day-to-day basis is because we are part of the community. That community is delivering value to us. Goes both ways, you know. And it’s the right thing to do to build value back to the community. That’s, that’s where we are, why we are very passionate, passionate about.

[01:05:05] Johan: So that’s why we also wanted to offer an open source version also, because not everyone wants to log in with their Google account and trust this. Yeah. You know, and also there’s, there’s companies that don’t want debt. It’s just a good thing to do. But the real goal is to set an industry standard and to do that, you cannot ask money for everything because no one will adopt, adopt.

[01:05:30] Matthew: I know you’ve kind of mentioned that you, you’ve, I don’t think I’ve really heard what you do or don’t do with, with sort of AI and l lms. I’ve heard fairly negative conversations about them so far. I think jumping out of a window, being one of them. But do you see other aspects of your, of your daily life where you do find utility in them?

[01:05:51] Matthew: And is it, oh yeah. Is it primarily in love? Is it primarily in the admin and the automation and and away from your, the technical side or how, how does that, how do you square up? 

[01:06:00] Johan: I use it every day, so don’t get me wrong, I’m not against it, I use it for other stuff, so I don’t, I use it for SQL code, code as well by the way.

[01:06:14] Johan: But we, when we are developing, typically don’t copy paste stuff from an LLM. We, because we have the know-how to build it ourselves, you know, we want it to be as good as possible. We don’t release to be the first to be fast. It has to be good and thoroughly tested. But I use it a lot for brainstorming.

[01:06:34] Johan: It’s, it’s brilliant. You know, I, I, I do mainly the marketing sales sites. I think a lot about price strategy, business. It’s all new for me and for me, that’s very, very exciting to not do the basics. Analytics engineering stuff, you know, I understand it and I do small things here and there, but I really chose to, in this company called Super Farm Labs.

[01:07:06] Johan: I chose to have another role and, and that is really, for me personally, it’s really refreshing and I, I’m chatting likes for hours, on a weekly basis basis, I’m probably chatting hours and hours, which they are t sometimes others just to test it. 

[01:07:28] Dara: Are you worried, I know this is maybe becoming a bit of a cliche question because everyone, you know, everyone either is or isn’t worried about whether AI is going to steal their living and take their job and make them redundant.

[01:07:40] Dara: You mentioned you don’t think the data engineering agent is good enough yet, and you don’t think it’s quite there where it’s trustworthy. Are you concerned about how quickly? It will get there. And whether there’ll be a day in the near term where somebody won’t need something like GA for data form and they can just go to their LLM of choice and say, here’s my credentials for my GA4 account.

[01:08:04] Dara: Build me a bunch of dashboards that help me make good decisions. And it just goes away, and does it. 

[01:08:10] Johan: So I’m not afraid at all. I’m more afraid, you know, of AI when it comes to ethical stuff, you know, that’s really something to worry about. I’m not afraid for my job to go away. ’cause if it does, then I will do something else.

[01:08:26] Johan: It will hopefully excite me just as much. You know, I’m, I’m very realistic ’cause I’m not saving lives here. And eventually we will probably all pivot to doing stuff that AI cannot replace. So if that’s the case, I’m up for that. 

[01:08:45] Matthew: Potato farmers and things like that. Yeah. 

[01:08:48] Johan: Yeah. Why not? I will probably be happier. Yes, that’s true. Yeah. So there’s only stuff to gain there. I think that in the near future, at least, Gemini is not optimized to make your Google Cloud bill decrease. Right. To, so yes, it can help you, it can probably automate everything for you, but you’ll, you’ll get a crazy invoice from Google that it optimizes. Yeah. That’s my idea. But I’m not sure. I’m just guessing like anyone else. 

[01:09:24] Dara: I know you said, just kind of steered a little bit back away from AI, at least for a minute. I know you said you’re focusing on the, you, you, you, you were kind of getting the basics right and making sure it’s providing value and you want it to be kind of an industry standard.

[01:09:40] Dara: But have you thought about it? Maybe expanding it beyond GA4 in the future so people could work with other data sources as well. 

[01:09:50] Johan: Yeah, so I mentioned the Google ads, the native data transfer that’s also also built in, you know, and now the query, there’s the Google Search Console exports, the bulk exports.

[01:10:00] Johan: We are thinking about the, the, those are the obvious ones. Then it becomes interesting because then you’re, you know, you’re, you’re really under the domain of the five trends and supermetrics of those, of this world. And, you know, we don’t want to do what they do ’cause why would we? That makes no sense.

[01:10:21] Johan: But it is interesting to think about, okay, we make sure you can plug in everything from Airbus or Five Trend into our models. That is a possibility. Yeah. So we’re working towards something modular. code. Yeah. That’s what we’re already doing. 

[01:10:42] Matthew: So if we really wanna know what your plans are, we just need to look at what de domains you’ve bought. This sounds like it’s a GA4 data form. If you’ve got all other stuff for data form, then we know that you’re planning on, expanding out. 

[01:10:55] Johan: Yeah. And then next week, five general buyers this year. That’s the end goal. 

[01:11:00] Dara: Well, listen, they’re buying everybody, so you’ve got a good chance of getting rich. After all,

[01:11:04] Johan: I will sign up for that every day, every year. You know, it’s, I don’t know, the, the getting rich is, you know, who, who doesn’t want to get rich? It’s not, it’s not a real goal. Whereas us it’s more about working together with a bunch of nerds Yeah. Who has been like, utterly rewarding already and we’re all doing this next to our normal job, so, yeah.

[01:11:28] Dara: Do, do you think that, you know, we’re kind of joking a little bit about five chan, but do you think if five chan is done by DBT, do you think that could actually work in your favor? Because it might. DBT will lose its kind of independent aspect then, and maybe that’ll push more people towards data form, which could potentially be good for you.

[01:11:47] Johan: I’m not sure. It’ll be really bad for DBT. Mm-hmm. That’s no question about that. Because what I said, it’s, it’s, it’s a very vibrant community market being acquired by like, like an enterprise company that’s probably going to close it, close gated. It’s not a good thing. But also worked at, like I said, I worked at companies where they were not friends of Google.

[01:12:16] Johan: Let’s put it in that way. And it’s a big step for companies. So if they are going to use data form, they basically have to migrate everything to Google Cloud. That’s what you’re saying. And that is a very, very big hurdle or a lot of IT teams. So if. Especially in enterprise. I, I would say the biggest part of the organizations are living in a Microsoft world.

[01:12:43] Johan: Some of them are living in an AWS world, and that, and a very small part is living in a GCP world. And, and that is not changing just because DBT is acquired by another company. Oh, data forms only, is only this thing into the cloud’s architecture. Yeah, no, 

[01:12:59] Matthew: I wonder if it’s an a, a data form to a certain extent is an archetype of what an acquisition of DBT would look like, because obviously before Google bought data form, it had a data form call.

[01:13:10] Matthew: You could open source it, you could do it solely where you wanted to, but slowly but surely, it’s just shrunk into being completely, tightly integrated with BigQuery and not having any of that other stuff. So it feels like it’s already happened once. DBT, like you say, is, it’s the, it’s the granddaddy of this stuff, isn’t it?

[01:13:27] Matthew: It’s the, the, it released the manifesto in the first instance about why we should be doing data in this way and, and all of that stuff. So it does feel like it would be a shame. 

[01:13:37] Johan: There will be someone forking Liberty, G Core. Right. And, and, and we free quickly and, and basically do what DBT does now. 

[01:13:46] Matthew: Yeah, yeah. Well, there’ll be a gap for it again, won’t then. Yeah, yeah.

[01:13:50] Johan: Yeah, definitely. 

[01:13:51] Matthew: so yeah, I ask this every fortnight, and sometimes it works and sometimes it doesn’t because sometimes we’ve been having long talks about, about AI and the future of ai, but essentially it is over the next sort of two or three years, where do you see things going?

[01:14:08] Matthew: Could be industry, could be with ai, could be, could be general trends. Like what, what do you think are the big changes that are coming our way over the next couple of years? 

[01:14:15] Johan: I don’t think I’m, I am thinking that much ahead because it’s probably smart. Yeah. It’s almost impossible. If you look at the speed of the movement in the industry, it’s almost unbearable.

[01:14:30] Johan: I, I, I just cannot imagine how it will look in two years. I can say stuff, but it doesn’t make sense. So I really don’t know. I really don’t know. And I mean it, when I say I really take it day by day, and I’m not married to Google or data form, I want to be agile. I even want to be agnostic, which is hard to do.

[01:14:53] Johan: You have a product that’s used for data form, you know, but I’m, I’m on a personal level. I’m transitioning more into, I would say, an entrepreneur. I want to start businesses and this is one of them. And other ones will be completely out of this data world. So if this is going nowhere or is going to be replaced, I will still be able to do business in some form. You know, that’s, that’s how I look at it. 

[01:15:26] Dara: That’s a good answer. And, and a good, yeah. And a good, and a good attitude to have, I think. Yeah. Very pragmatic. Probably the only answer to this question is that we don’t have to look back on it and see if it was right or not, because I think a lot of people, they do speculate, and I’m not, I’m not knocking now, that’s why we asked the question and we speculate as well.

[01:15:44] Dara: But I think you’re right. It’s impossible to know that it’s changing so much at the moment that, you know, it’s not like five years ago, you could have predicted what was going to happen in analytics 12 months later or two years later or whatever. But it’s all changed now. and I don’t think anybody really knows.

[01:15:59] Johan: No. And if you, you know, if you look at yours, what I worry about is, as you, as I said, I come from a journalist background that supports I’m, I worry about, you know, yeah. We are almost at a point where there’s no way to see if something is real or fake. That stuff is scary. 

[01:16:17] Matthew: I mean, I feel, I feel, I, I think I was saying this, we had a, an old company sort of hackathon, last week, and I was, I was talking about Sora too, with a couple of people.

[01:16:27] Matthew: Yeah. And there were a few videos that came out of that. I couldn’t tell if they were real or not. And then it was a strange sensation for the rest of the day. Whenever I saw a video and things, there was just this bit in the back of my head saying, is this real? And I, that was pretty terrifying to think that that is, especially as people are already being convinced by things that to many people would be ridiculous, but it looks almost indistinguishable or completely indistinguishable from real life. I mean, what, what the hell happened to them? 

[01:16:57] Johan: Indeed. So that’s like, yeah, I think in two years we could be living in a Black Mirror episode, you know, or caves, but let’s not end this. Put that with, or we could 

[01:17:10] Matthew: be living in a utopia. That’s the, yeah. The positive way to end it 

[01:17:14] Johan: with a positive way. Yeah. All thanks to G four data form.

[01:17:18] Dara: Absolutely, yes. Alright, Johan, thank you very much again for, for coming back on and, and updating us on, on what you’re up to these days. And maybe in another 75 or so episodes we’ll bring you on and you’ll have the, the next company that you’ve, you’ve set up and we can talk to you about that. 

[01:17:35] Johan: It was my pleasure. And, next time I will send my AI agent. 

[01:17:39] Dara: Yes, we all, we all will. Yeah. It looks good. Alright, so thanks again, Johan. Thanks. Bye-bye. That’s it for this week’s episode of the Measure Pods. We hope you enjoyed it and picked up something useful along the way. If you haven’t already, make sure to subscribe on whatever platform you’re listening on so you don’t miss future episodes.

[01:17:57] Matthew: And if you’re enjoying the show, we’d really appreciate it if you left us a quick review. It really helps more people discover the pod and keeps us motivated to bring back more. So thanks for listening and we’ll catch you next time.

• Spotify
• YouTube
• RSS Feed