#97 The future of event data (with Timo Dechau @ deepskydata)

Written by | February 9, 2024
The Measure Pod
The Measure Pod
#97 The future of event data (with Timo Dechau @ deepskydata)
/

In this week’s episode of The Measure Pod we spoke with Timo Dechau, a product manager and analyst with a unique perspective on event tracking. He discusses the challenges faced by product managers, his journey into product management, and the technical skills he deploys. We spoke about event data and marketing data, as well as the people, processes, and design involved in event tracking.

Show note links:

🎥 The podcast is now available in vodcast (video) format! Watch the episode below, or over on YouTube.

Let us know what you think and fill out the Feedback Form, or email podcast@measurelab.co.uk to drop Dan, Dara and Bhav a message directly.

Follow Measurelab on LinkedIn and on Twitter/X, and join the CRAP Talks Slack community.

Find out when the next CRAP Talks event is happening on LinkedIn.

Music composed by Confidential – check out their lo-fi beats on Spotify.

Master Google Analytics 4 with Daniel Perry-Reed on the next GA4 Immersion 6-week cohort training course. Charity and group discounts available!

Quotes of the Episode:

  1. “…And product is more complicated because like there’s often not like a clear funnel in there, it’s not a linear path that you have.” – Timo 
  2. “…”It’s the irony, isn’t it? Is that people think that it used to be okay and now it’s broken. And actually it was always broken.” – Dan

Transcript

Intro | Topic | Rapid fire

Intro

[00:00:00] Dan: Welcome back to The Measure Pod. In this episode, we have a really special guest, Timo Dechau. And I’ve been a fan of his work for a long time. And actually Bhav we discussed it in the episode, but one of our previous episodes where you were guest host in the last season was actually spurred off of the back of one of these articles he wrote on Substack.

[00:00:30] Dan: But yeah, I mean, this conversation went in a lot of different directions. We talked about event data, marketing data, user level data, everything. I just found it such a fascinating insight into that world that we all work in, and his perspective is just really, I don’t know, enlightening, refreshing.

[00:00:45] Bhav: I’ve got a proper buzz from that conversation, I really enjoyed it. I mean, with me, he’s definitely like preaching to the choir with the things he talks about, but it was so nice to go into some details. You know, of course we talked about some of the more technical elements, but I think what I liked about Timo is that A, his background, he comes from a product background, which means he kind of has an innate understanding of what challenges are out there for product managers.

[00:01:08] Bhav: But then as an analyst and, and a technical person, he’s now being able to like bring those worlds together. And you rarely see that kind of complementary skill sets come together, and it was really great to hear about, you know, the people, the processes, the, you know, the designs of event tracking. I think it’s yeah, it’s been a great episode, and yeah, I think we should just move out of the way and let people listen to it.

[00:01:31] Dan: For sure I do want to address the fact that as soon as you said his background, I did think you mean his visual background. So if you’re watching this on YouTube, I think he’s got like different LEGO sets behind him all set up. I think it was LEGO, anyway, there was some really interesting bits and bobs in the background, but no, you’re right.

[00:01:44] Dan: You’re right, we’ll keep it all professional. Yeah definitely not the background off the camera. Well, just a couple, a couple of plugs from us. As ever, we’re audio and visual format now. So you can catch us on YouTube or in any of your podcasting apps. We also hang out on the CRAP Slack channel. So if you want to come chat to me, Bhav and other communities, like other people in the conversion rate analytics and product world, you can find a link to that, all of these things in the show notes. And I think that’s pretty much it. Anything else to plug Bhav before we jump in? 

[00:02:11] Bhav: No, it’s been a great episode. I just hope everyone enjoys listening to it as much as we’ve enjoyed recording it. 

[00:02:16] Dan: Amazing, all right. Well, enjoy the show. Today we’re joined by Timo Dechau and I have been a fan of your work for a long time, Timo.

[00:02:26] Dan: We’ve actually had you as the subject of the MeasurePod before way back in episode 84, but this is the first time you’re here. So first of all, welcome. And thanks for coming on the show. 

[00:02:34] Timo: Oh, thank you for having me. And yeah, I can still remember it. And I was listening to it and this was, it was a weird experience because like I was listening to the episode and I was always wanting to jump in. I was like, these are really great points. I want to extend on that. And so it was like, but I couldn’t, so I could just talk to myself while I was driving. So it was nice. 

[00:02:53] Dan: Well, now you’re here and you can defend yourself. If there’s anything you feel like you need to, you can, of course. Well look, Timo as our frequent listeners might know, is we don’t introduce our own guests. We let our guests do that for themselves. So Timo, tell us the origin story of Timo. How did you get to a point where you’re talking to people like me about analytics on this podcast?

Topic

[00:03:09] Timo: Yeah, so I’ll try to make it not too long. So I started out in product. It’s also like explains why, why I’m deep in product analytics still. And so the reason for this is like when I was starting out in product and it was really a long time ago I really had a hard time to basically. So put it like this I had a very conservative kind of company and other people were running my backlog. So like all the other teams will basically run my backlog. I was in product, but I was basically project management. I wanted to do more things that I think were good for the product and I was looking for ways how to do that.

[00:03:48] Timo: And since I always had an affinity for data, I thought maybe this could be something, and it was really like the early days. So this was early days, even GA early days. It was the old Google Analytics classic, where you, where you didn’t even have segments, so you could create profiles and then wait two weeks until, or some days, not two weeks, but some days until the data was coming in. So, even crazy time from an analytics kind of standpoint. 

[00:04:16] Timo: But it definitely helped me to change agendas. So it definitely helped me to track some things and to get some insights into, let’s say usage of specific kind of features and other things, to then get people to say, yeah, maybe it would be helpful to go for this kind of improvement and so on.

[00:04:34] Timo: So it opened up a new road to talk to people about, okay, how can we improve products? And I was still around in product like this for six, seven more years. But data was always like an important part, wherever I started out new, most of the time we didn’t really have a good analytics setup. So this was always like my first part of my job to do this, to enable this, because like I wanted to have this fast feedback loop that you can get from working with event data to iterate on product features. And then at some point I basically just did this as my main job. I got a little bit fed up with working in product and I thought, okay, I need a new perspective.

[00:05:13] Timo: And I just picked the one thing that I always did and just basically the data and then also like broaden the thing. So I basically left a little bit product analytics, did more marketing analytics cases. Then even did BI cases of building up data stacks. And then, you know, I would say in the last two or three years coming back to the, to my roots and really like focusing a lot on event data, because like, I don’t know, I still think it’s the best kind of data to do stuff with.

[00:05:43] Timo: So I mean, this is definitely arguable. I mean, I’m in a good round here so I think it might be not so many people have, let’s say strong counters, but yeah, I still think it’s, it’s the best way to work with data. So, and I now try to get more people on board yeah. 

[00:05:58] Bhav: So as I was saying, I think you’re preaching to the converted here anyway at least certainly on this, on this call. I want to just mention one, you know, you said the early days of your product career the other teams were pretty much running your backlog. I don’t think that’s changed in many organisations still. I still feel like there are many product organisations where the product teams are run by, you know, they have a product manager who’s, who’s the face of that, of the product, but it’s still pretty much run by other people so. 

[00:06:24] Bhav: But one thing that definitely cuts you apart from the rest of the product folks, and certainly many of the product folks I’ve worked with, is that realisation on the importance of event data, because it’s something that gets overlooked. So many times I’ve lost count and so it’s refreshing to hear from a product perspective, you know, someone who’s sitting there talking about the importance of event data so much to the point that you’re actually writing about it. 

[00:06:50] Dan: I mean, that’s interesting. So my background is marketing analytics and still is. And so that’s where I’ve kind of carved my career path into, and you know, the move to things like Google Analytics 4 and throwing or thrusting event data onto a lot of traditional marketers that are session based and hit based or page view based is, is a really interesting perspective and it’s a really interesting transition because I’m very aware of the wider industry in the product space as well, working with Bhav as well, in terms of that GA is just playing catch up to every other tool out there. 

[00:07:20] Dan: And actually it’s just a laggard in that aspect in terms of catching up. But at the same time, you’ve got this whole kind of cohort of people that are in the marketing space that are just like completely bemused by this move and not understanding the reasoning behind it or understanding how to use it. But it’s interesting to see how different the industry can be, but at the same time, I don’t know, we will still talk about the same stuff in the same way, we just have different words for stuff right. 

[00:07:42] Bhav: I was going to say the point around, you made around page views being kind of like the, the classic way of measuring web analytics back, you know, way back when I think you were ahead of the technology back then and you having to leave and, you know, do some marketing analysis maybe and now coming back into product analytics over the last few years is probably the right time because I feel like the technology is just catching up in the sense that it was probably always available, but it’s now widely accessible in terms of the cost factor having come down. 

[00:08:13] Bhav: So more and more companies can now think about product analytics and sending an absolute abundance of data to their data warehouse or to some, to some platforms. So I think now’s probably the time. And you know, most companies still haven’t adopted this product analytics mindset. I think if you look at most e-commerce companies, they still maybe operate in that traditional page views, web analyst way of thinking from 10 years ago. 

[00:08:39] Timo: I think it’s true. So, one thing about page views, I mean, page views in the end are also events. I mean, we always had events in this. So, like, technically, they are just events so. But I think it’s right, I think a lot of things have changed and it’s quite interesting. I mean, it’s also interesting how the whole product analytics space was basically even created. I mean, I would still say it was heavily influenced by mobile a lot because like, I mean, the, the big player, like Amplitude and Mixpanel where before mostly mobile analytics and then Firebase came around and basically destroyed the category. And not, not only destroy it so I think there are several reasons why this category never really took off. I think this is a podcast by its own. 

[00:09:16] Timo: But I would say working with apps already introduced. You had to introduce a new mindset because like, it was not really working like this. So you didn’t really have this clear six step funnel that you can still track, maybe just with a page view or so on. And so apps were more complicated. I think this started to trickle these things. And also, like, I mean, this is the same with GA’s journey. So GA started out, or let’s say that what we have now GA4 started out on the mobile side and then obviously came over to the web or to just web. I still don’t get why GA is not good in doing server-side stuff, but yeah, this is also like a different discussion, it’s stupid. 

[00:09:55] Dan: Don’t ever question Google on these things because I don’t think they even have an answer. And if they did, I’m sure it’s not going to be a good 

[00:10:02] Timo: It’s something no one actually wants to hear. So this is something else, but you’re totally right. That at least like, I think today we definitely have a different kind of setup. And so it’s still hard to get product teams to work with event data. I still have the struggle today and I don’t really have a clear answer to that.

[00:10:21] Timo: I don’t always want to say it’s quite hard because this maybe demotivates people, but it’s often not so clear as you have it in marketing analytics. I think marketing analytics is something that you can teach people easier because like you can start with, let’s say very simple things. So like, okay, you send a lot of campaigns on your website and then at some point you want to have some kind of success or some conversion events. So this is the simple thing, like simple funnel to events. Someone comes on the website converts and then you can build from there. You can say, okay, look, let’s maybe put some more details so we expand the funnel and we have some more steps. 

[00:10:56] Timo: And product is more complicated because like there’s often not like a clear funnel in there. So it’s not a linear path that you have. I mean, if you have a software as a service, you want to get the people into subscription at some point, but the way to a subscription can be totally chaotic when you just look on different kinds of, let’s say, journeys that users have, and then to develop these kinds of tools to either go on a very high level or to go really deep down to identify this kind of patterns where you can then say, okay, look this could be interesting that we can push people in this direction. And we can analyse it with data if they go, for example, if we get them to experience the first value from a product, and then we figure out how to measure it and then we measure it.

[00:11:38] Timo: And then we can run a lot of iterations and optimization also with customer success and with growth to get more people in this kind of phase. I think this still takes a lot more understanding, first on the data side, which is not easy. And then second of all interdisciplinary. So you have to work a lot then with the other teams to understand what they are doing and so on to bring it all together. I mean, this is still a part of which I love about this thing, which I still like to do this, I think, which why I came back because like you can do really crazy stuff and you can really identify some interesting patterns when you work with the data, which still makes it quite exciting to work in this kind of way. 

[00:12:20] Bhav: I think one of the main reasons for this and, and I’d love to hear your thoughts around why you think it’s like this, because I think your point around marketing is a bit more established, it’s easily understood. And I think a core part of this is because of the fact that it is so standardised.

[00:12:35] Bhav: You have, you know, your impressions, you have your clicks, you have your landing page views and your bounce rates and then you have your, you know, like your conversions and your spend. And I think it’s such a standard model that it can be adopted quite easily. There may be some like permutations and some nuances, but by and large, it’s a very standardised model. I think when you enter into this product world, each product is so different from another product. And it’s, you know, you have your SaaS, you have your services, you have apps and I think one of the reasons why it’s so difficult and because it is, it is difficult just by itself.

[00:13:10] Bhav: One of the main challenges I think is because the way that it’s perceived in an organisation is that it’s an input first focus instead of an outcome first focus. And I think if we switch the model to say, what is it we want to learn and then work backwards from what events need to be implemented to be able to learn that, I think that we, you know, we could potentially see a shift in mindset. So I’d love to hear your thoughts around what, you know, why you think it’s so challenging. 

[00:13:38] Timo: The starting point I think is really essential. So I was trying to do product analytics training with product teams for some years, and I have to say I was highly unsuccessful. So it’s like, I would not say I wasted people’s time, but I really had always this experience when I work with, let’s say 20 people do the training. It was usually one or two people who really benefited from that, the rest basically stopped listening after 10 minutes. When I was, when I was coming up with, okay, let’s do some cohort analysis, basically I lost the room.

[00:14:06] Timo: And because this was, this was my mistake because like, I think I took it from the wrong, wrong side because you’re right. So yeah it’s too abstract. So it’s not really bound to how they work every day. So one thing that definitely is helpful, and I think this is also like a good point, if you want to get started with product analytics, it’s like to go from the features because a feature is like something that aligns a lot of, also like it aligns in the company, more people.

[00:14:30] Timo: So it aligns the developers who are working on this and the developers, we all know, play a very important part in the whole data set up as well. And then it aligns, of course, like the product team, but it can also align then a customer success or, or gross team because like they might want to drive, I’d say feature adoption, or I want to use the feature to improve let’s say the growth of accounts.

[00:14:52] Timo: And so it brings a lot more people on the table. And when you go from a feature perspective and you say, okay, look, okay, why are we building this feature? So then you might struggle at still there are product teams around there that don’t see features as a bet and still think it’s like, it’s given that this stuff will work, which it’s not. I mean, no feature actually has a guarantee to work. 

[00:15:12] Timo: But it’s much easier to establish this, like, okay, look, you introduce this feature, let’s figure out what you want to achieve with that. And this is something where at least like you can, you can bring 10 people in the room and you will have a good conversation about that, where they think, okay, we, how we improve this. And then you can help them to break it down into some kind of metrics to say, okay, well, let’s say, okay, we want to achieve this, let’s break it down into three, four metrics that can help us to get us, let’s say some, this quick feedback, if we are actually going the right direction. And then from the metrics, you can just then define the events that you need, then you implement them. 

[00:15:46] Timo: And then the whole thing totally makes sense. And then in the end, like you can even, you even know what you have to implement then let’s say in some kind of reporting before you launch the feature, before you deploy the feature. And so then you make it basically a company exercise once the stuff is launched to actually just check the feature dashboard or report or whatever you have to always keep looking, okay, how do we do, how do we make this? And at least like I did this two or three times in product teams and you can create a lot of motivation around that. 

[00:16:17] Timo: So like everyone loves to see this stuff and it also helps you a lot better to communicate this internally when you have something where people like, hey, we launched and now you can see live how the feature adoption goes on and maybe you roll it out with feature flags and you say, okay, now we have 5 percent, looks really good, we can dial up to 10 percent and so on. So if you establish this kind of work, then product analytics becomes more approachable because like, then it’s not so complicated anymore. Then you leave the other stuff maybe for the analysts to make really this deep dives, but yeah, but you’re totally right if you find some good use cases where teams are actually working on every day, and then you derive the data from there, it makes it a lot easier to go from there. 

[00:17:00] Dan: So this is a, this is like a universal thing is how do we make data profitable? Like, how do we get value from data? Because I think you can, you can interchange any you know, any service, any kind of area. I mean, marketing analytics, product analytics, gaming, whatever, like it’s about, can we make an analytics function, agency person, a profit centre rather than a cost centre, right. And that’s basically what we’re talking about here and tying it directly to kind of values features to kind of what the hell are you going to do with it if we gave it to you, I mean, establishing that is a good rule of thumb first. 

[00:17:30] Dan: But I want to kind of pull it back slightly, Timo, and just ask about like, so if anyone’s listening or watching that, they know the title of this episode is going to be the future of event data right. And so, this is spurred off a conversation off the back of one of your posts on Substack that I’ll link in the show notes for everyone else. 

[00:17:45] Dan: But I’m really interested to get your perspective on this so far has been around, like focusing on like how to get the value and how to get an adoption of event based data and event schemas for product teams or marketing teams. I mean, where does like, what’s the future of that? I mean, if we’re talking about getting adoption in event data now, like what’s next? I mean, maybe introduce what you’re kind of thinking about in terms of the future event data and does it kind of fix this problem or is it slightly different? 

[00:18:08] Timo: Technology never fixes problems. But I think that there are two things which are interesting for the future. So first of all, it’s like, I think the analyzers of sequences was something that at least like in the global data space, never really happened. So BI data is often like a snapshot data. You take a snapshot of your business data, of your sales data, put it in a dashboard. I mean, of course, like you have a time series and you can see, okay, our sales developed like this, but you don’t really, you don’t really analyse the sequence of stuff that happens in the business, how they enter, how they relate to each other, how they influence each other and so on.

[00:18:39] Timo: So I think sequence analysis is something extremely powerful. It’s something that people who started out on the analytics part immediately get, because like they, they basically work with this a lot of time. But I think like when we, when we apply to more and more data use cases across the company I think we can establish really interesting things. So for example we run more and more complex systems under the hood to run the stuff that we do. And all these systems are basically event based most of the time, even most like today, even like by paradigm when, where they’re using event driven kind of architecture, so they have, I don’t know, Kafka or other kind of queues in between where they’re just sending events in there. 

[00:19:18] Timo: And then they have subscribers to the event that they do stuff, which is a great breeding ground for, let’s say, getting event data out of the system, then analyzes. You have the possibility to analyse how your system is working and so on. And so the more and more independence, we want to put on our systems we have to have ways to look, still look into this and to analyse how they work. And so this is, for example, what you can do also like with event data. And I think this comes to the second point is the setups that we have today and the possibilities and also like driven what we did on the data stacks in the last five, six years, we now have possibilities to source events from a lot more sources.

[00:19:56] Timo: I mean, one of the problems was event data was always instrumentation. So first of all, development team never really wanted to do instrumentation. It took a lot of time, then it was very error prone because like, it’s easy to get wrong, but now we have, basically we have a lot of ways to get this stuff. So we can just hook in and our production databases and just pull events directly from there. I don’t want to say, okay, you have to do it by CDC. I think this is also like a little bit overhyped right now. But CDC could be like something, so change data capture. 

[00:20:28] Timo: So where we in the end just look into production database, look, okay, what has changed, has something created, has something updated, and we derive events from that, or what I often do is like, I can basically get a lot of event data directly from webhooks from third party systems. So if something is happening in your CRM or something happens in Stripe, on a subscription and so on, I can just derive a lot of events from there. So in the end, like when I did one or two setups now where I was surface, or let’s say sourcing events, I think at least for 80%, not from any kind of tracking anymore.

[00:21:03] Timo: So I think tracking in the end was just 20 percent of that. And I could get it from all the different kinds of other systems and this enables more use cases. And then now we have, I think we are in the first time where we have also like the tools that can build on top of this data. So then I can put all the event data and data warehouse and I know you did the episode with Adam where you’re talking about this. So now you can put the event data and data warehouse. And this again introduces a new way to control this kind of data. 

[00:21:33] Timo: So you can basically have a raw event layer where you can pump any kind of weird event data in there where everyone would say, if you do it directly in an analytics system, it’s too noisy, but since you put it in a data warehouse, you can then control it so you can merge it. You can filter it out, you can just say, okay, I just make this 20 percent visible to my users because like you basically model it on a specific kind of layer and then you just pop something, analytics tool, Mixpanel, Amplitude on top of that and work with it and then give it to the users. 

[00:22:06] Timo: Also like it, it helps you to qualify all that to increase the quality of the data significantly, since you can change the data before it goes into the analytics tools and so on. So to improve something, which I always call like the data user experience, which is always hard to do when you pipe that data directly into the analytics system, because we all know like the event is wrong in Google Analytics, at least in the old one. And the new one is a little bit better. 

[00:22:31] Timo: You have it there forever, so it’s really like, you cannot really get rid of it. But now we for the first time, I think we have the possibilities to put something before that, which can control, which then gets in the hands of the users, and I think this makes them also like the adoption easier because like people don’t really have to work around the weirdness that has already gone into the system. I think these are the two major things that I see for the future for event analytics. And we are just at the starting point, so we don’t really know how it plays out, but yeah I would do a lot of stuff this year around this, especially like about weird cases, which no one really thought about what you can do with event data. 

[00:23:09] Timo: So one thing that I definitely want to do is like collecting all the metadata of data pipelines and just pop this into an event analytics system to just analyse data pipelines performance with that, because in the end they, they do the same thing it’s all event based. It’s just not really visible by its initial shape, but you can easily shape it when it’s just event based. 

[00:23:32] Bhav: Let’s pick that part a little by little if we can, if that’s okay. So we start off with, you initially kicked that off with a shot towards like the BI and the fact that BI is a snapshot in time. And it’s even with the time series, it’s still just it still doesn’t give you everything we need. The evolution of BI though, is that going to be driven through product analytics in the sense that we’re just reporting more numbers that are event based or do we need to move away from this concept of BI? And like do we still need people like view, visualising a million metrics on a dashboard?

[00:24:10] Bhav: At what point does this stop becoming useful? Then you kind of moved into the fact that you’ve, a lot of data, certainly for yourself, comes from non traditional product analytics, data capture methods. 

[00:24:22] Timo: I can add two things because like you brought up two, two really good points. I think we will still have dashboards. So I think like dashboard has definitely its purpose and its role. I think it was just over purposed. I think like we didn’t really had a good answer to things. And when we don’t have a good answer, we take the next big thing. And this was, this was always like dashboards and so we were trying to do a lot of things. 

[00:24:41] Timo: I think of course, and you can already see the trend. Dashboards becomes more simple, more focused. I think people are getting better at really figuring out what are the core metrics they need to report on. And I think this is like where dashboards are still good.

[00:24:53] Timo: I mean, I still use dashboards because like you have one glance, you can really see, okay, something is off or not and I think this is still valuable. I think the interesting part that we can introduce with let’s say event data and sequences is like, you can answer different kinds of questions. So Ahmed from the Rater who I talk to a lot and who’s, he can do even better monologues on this than I can do. He always has a good set of questions where he said like, this is something you cannot answer in BI, but you want to answer. 

[00:25:19] Timo: So like okay look, we rolled out this new product, after this new product has been rolled out we want to analyse how many people are basically watched all the details about this product and then bought it and then basically we’re in touch with customer support and then how many of them are still retaining on our website or on a shop for the next 12 months, these are complex questions, but in the end, like these are questions, which are in the heads of product teams or so on, and you cannot answer them with BI easily. 

[00:25:46] Timo: So if you want to answer them, it basically takes you weeks to prepare the data and to get, and this is something where a sequence analysis really shines because like in the end is what I just described is a sequence of events that are happening somewhere. And so we just have to bring them in the right order and then start to analyse it. 

[00:26:03] Bhav: Yeah, and I agree. And I think it leads to kind of like the next thing you talked about, which was when you have a product, you know, in order to understand what’s happening, you kind of break it down into, well, this is the outcome we’re trying to get to, what are going to be the inputs, and then you define the events off the back of that.

[00:26:17] Bhav: I think you’re describing what I call a metric tree or a measurement or some type of measurement framework. I live and breathe these, I’m a big advocate for them, I write about them a fair bit, so. I think this is the critical element that’s missing, people jump from BI to product analysis without taking that kind of like necessary step in between where we’re like, hang on guys, let’s stop. Before we throw out a million events that we’re going to start tracking and build a million dashboards off the back of it. What are the things that we really care about? And I think that becomes a starting point around making sure that your BI stays relevant and you track the right things.

[00:26:53] Bhav: Then you said something really interesting and I never really thought about this. You talked about webhooks and you talked about the fact that 80 percent of your data from product analytics perspective comes via webhooks and actually not needing to integrate and do the product analyst and the event tracking needed in those platforms and those third parties. You just send some type of event into your product analytics platform when something happens. Do you think that’s scalable? Like, because most people don’t have that much third party. Maybe I’m wrong, but is this going to be something that is widely adoptable? 

[00:27:29] Timo: I think it is. It’s also not always coming down to webhooks, webhooks is one way, but you can also like, so for example Stripe is a good example. So you can pull Stripe data with any kind of the obvious let’s say data integration tools. So for example, if you use airbyte, Stripe has an extensive event API where you can pull all the events that are happening on Stripe.

[00:27:51] Timo: Not all integrations are supporting it. So for example, I think fivetran is not supporting it, but if you use airbyte, you get the event table. So you basically get a raw immutable event table, all the stuff that happening in your Stripe account. And so you don’t even, you have to use webhooks there. So you can just pull the data in a nightly job down and you have event data and you can also pull other data down and basically eventify it. So, because like some data, even when it’s snapshot data, you can still make events out of it. And by that I get to the 80%. 

[00:28:19] Timo: The interesting thing is what you said with webhooks is I think teams often underestimate how many tools they use on this kind of thing. When I do sessions with team, where we figure out how to do an event design, one of the questions is like, okay, what are the kind of tools are we using to serve your product in the end? And people are starting out with yeah we just have three or four. We always end up with 15, 20 because like you have everyone in the room, then everyone’s like, yeah, but actually for this onboarding sequence, we use XY that. 

[00:28:49] Timo: And so you often really have a lot of these and often these tools are pretty good in sending out this kind of, and not every tool is supporting webhooks, but a lot of tools do. And so, especially like, for example, CRMs. You don’t really get easily event data out of HubSpot. I think you can do it on the enterprise plan, not so many people spend so much money on that, but HubSpot is sending events around everything they do. And so in the end, it’s just like ask IT to set up a proxy that can receive the stuff and then do whatever you want to do with it. 

[00:29:24] Timo: So either write it directly into data warehouse or pass it on directly to product analytics tool. And so I built these proxies now a lot and so I still have some running around for myself and they are pretty straightforward, simple tech. It’s not so hard to do, it’s pretty bulletproof because like, there’s not so many complex things happening. You can just scale it on one of the very good scalable services and yeah, and it works. So it’s a good way. 

[00:29:50] Dan: I do a lot of similar things in the, the kind of marketing analytics space right. So you mentioned things like the kind of maps where they kind of escalates out of people think they’re spending in three platforms and, you know, it turns out that they’ve got like all these different sort of tools, they’re all connected to each other in lots of different ways and doing some kind of visual map is always a really interesting way of putting it in front of them. And they’re like, oh crap, we’ve got a lot of stuff we need to consider. 

[00:30:10] Dan: But again, I can only speak to my own experience and a lot in the, in the kind of marketing space, there’s huge changes and movements towards kind of like not providing event level data, right? So we’ve had to stuff like for example, the data you get out of things like Google Ads and Facebook Ads, they’re all going to be aggregate data sets and actually there’s no way to get event level data in there.

[00:30:28] Dan: So where does data like that sit into this kind of ecosystem and I mean, you know, we’re building kind of warehouses and we’re using techniques like media mix modelling to kind of understand things like uplift, doing experimentation in marketing campaigns to understand influence. And, you know, back in the day, you know, in the wild west of the days that I think we all here remember, it’s like you tracked everything and you had all the events and you can do the kind of attribution modelling, but a lot of the data I play with is becoming aggregated. And so I think this event data is a pipe dream and maybe it’s going to become sort of having diminishing returns over time. So what happens there? Like in those kinds of situations. 

[00:31:04] Timo: A really good point. And definitely not easy to answer. I think one thing that what we, I think we might see is like that. I mean, you already see it on the big platforms. It’s like you move the point where you can identify people and you basically own the people. So in the end, it will become a business strategy to own more of the people’s journey or more of the people’s touchpoint on your systems and not on external systems. So like, I mean, maybe we see more in e-commerce that people are basically, I mean, they are definitely tendencies. 

[00:31:38] Timo: So like all the things to get people to sign up for a newsletter early on. Of course it’s for the newsletter, but it’s also like for the data side to identify them as early as possible or re identify them. And I think this, so it might be like that we see models that are trying to get people into signing in much earlier to do these things. From the advertisement side, you will still have the aggregated data and I think this is like, and also like you will have also like from the front end side, you will have less data this is the other problem. 

[00:32:06] Timo: Based on consent you basically will not have so many user based data anymore. And so this, of course, like makes the other part more interesting where you, let’s say, pull more system like data from the other kinds of systems, but it doesn’t make the topic of marketing attribution more easy. I think the topic of marketing attribution is definitely something. Okay I think it was a pipe dream 10 years ago where people thought, okay, we track everything and now attribution works just fine.

[00:32:37] Dan: It’s the irony, isn’t it? Is that people think that it used to be okay. And now it’s broken and actually it was always broken. And I find you know, there’s a, yeah, I find a lot of the stuff with the, you know, in my world, moving from Universal Analytics to Google Analytics 4 people assume that Universal Analytics was a source of truth and perfect in every way. And then you’d kind of, you know, I do a lot of training in that aspect and I explained to them what bounce rate actually means for example, they’re like, wait, what? I was using that in my dashboards and I didn’t know what it meant. And it’s opening up a lot of stuff in that context. 

[00:33:05] Dan: And so I get that and I think, you know, owning this kind of first party data set and kind of joining what you can and owning more of the journey and we’re seeing, I think everyone is aware of all the kind of, not just the banners for consent, but also things like login portals and I suppose once you get behind into a product, it becomes a lot easier because they’re going to be signed in or identified in some way, but there’s going to be a lot of data whereby you’re not going to be able to join it together. 

[00:33:27] Dan: There’s something else you wrote about Timo, which actually I’ve put in the show notes and I don’t want to turn this into a CDP based conversation. But what we’re talking about here is a really important part of this, which is collecting data and doing identity resolution and joining the dots based on this matrix of IDs, because every system is going to have their own IDs and you’re going to have this task of, you know, email address, this ID, HubSpot ID, Google Analytics ID. So god, that must be a task in itself. Is it as hard as it sounds? Or is this just a chore you have to get through or is there ways of kind of addressing that in a more streamlined or sensible way? 

[00:34:00] Timo: Interestingly, it’s a topic no one actually ever talks about, but I would say this is the crucial thing. If you do any kind of customer journey data analysis or customer journey work, which in the end like comes down to, I don’t know, do CDP stuff. I still, so I had a time where I worked a lot with B2B software as a service companies and everyone had the same problem. No one could basically create a funnel of the whole customer journey until someone, let’s say is set up and up and ready that they can use the tool until this points from the first touch point, everyone was struggling. And the reason was okay, to some degree let’s say maybe front end let’s say problems.

[00:34:38] Timo: But the other problem was like the whole journey went across different kinds of systems. So they were starting out on the website and they were signing up. Sign up was usually on a different kind of system. So like the application, which always people tell then maybe it creates an account, but still like it moves onto a CRM and then maybe like another team is taking over this kind of person that qualifying that they’re helping them and at some point they come back to the application at some point, magically in another system, it creates a subscription.

[00:35:04] Timo: And of course, like you have a lot of let’s say breakpoints on this whole journey and stitching this together is definitely not easy, but it’s the only way to do it. So it’s really like, it’s a hard exercise, but it’s something that you really have to do, but it’s definitely doable. So, I mean, as long as you can control everything, you can basically pass on identifiers to every kind of a next step that you do.

[00:35:30] Bhav: That statement, as long as you can control everything, I’m going to get that printed on a t-shirt. 

[00:35:37] Timo: No, it’s totally true. It would be really good t-shirt, but also like a scary one. I think you scare people off when you, when you have this, but in the end, like it already tells you what you have to do on the implementation side. So you have to take control over the different kinds of touch points if you want to have a good data set up. If you push a lot of things across different kinds of platforms and you pick tools, not by checking, okay, how much control do I have with the input that I bring in and you basically just rely what they do, it’s almost often impossible to basically build proper data journeys with that. 

[00:36:12] Dan: So you’ve mentioned a couple of times, like, as long as you take control, you own this journey. Who’s you in this conversation? Is that existing team structures? Is that BI? Is that product? Is there a dedicated analytics function? It all sounds amazing and perfect, but I’m struggling to think about who owns this process or is it something new that has to kind of evolve out of archaic on kind of existing structures?

[00:36:35] Timo: I think first step is like that the company owns it. So even this is not guaranteed. So I think this is step one. I think in the end, like this is cross owned, so this is a cross ownership. So you, for example, you can have a technical ownership. So this will be with development, but you can have, let’s say a operational ownership that might sit somewhere else. I would still think the data team is the best team where this can be owned because like the data team in the end, like has the ideas, what they want to do, what they can enable with this kind of data.

[00:37:04] Timo: But it requires a data team that is extremely proactive, that extremely well in understanding the business. Not the business rules, but the way how, for example, marketing teams are working at the moment. So how the product team is working, where are they challenging? Where can it support them? So it has to be a data team that identifies, for example, that the sales team really has problems to prioritise which kind of accounts they should focus on.

[00:37:29] Timo: And so when they understand this, they can come around and say, look, it’s pretty easy for us to make, to collect some behavioural data, maybe to develop a score for you to make a better decision, who should contact first and then second and so on. But this requires that the data team becomes really proactively involved in what the other teams are doing.

[00:37:47] Timo: And then proactively driving this technical side as well, to go to development, to go to marketing team and say, oh, look, we introduced a new CRM or we introduced a new email team to say, look, okay, let’s figure out how we can support you that you can get all the data stuff out there of this tool that you need to do a really good job. And so it’s not easy to do this, but I think still it comes down to the, to the data team, because like they usually, at least they usually have the technical tools, the technical skills. I think the only thing what they maybe have to improve is like this connect with the, with the business teams too. 

[00:38:26] Bhav: I think because we’re on the subject of people, I want to just bring up the challenge when you’re trying to scale something like event tracking. Because up until now, event tracking, I say up until now, I’m not really speaking to any specific timelines, but there was the previous period in time where the people did minimal event tracking, or they did event tracking without realising it because it was done out of the box, like page views and what have you.

[00:38:49] Bhav: Then it became a very conscious action where people were starting to track events, but it was never something that was strategically thought about. You mentioned in one of your blog posts I was reading that obviously as people start to add more events and you haven’t got the definitions in place, people start to add more and more events and then that starts to escalate, and what you get is this snowball effect. And I think my one big worry about product analysis, and I bring this up because we’re talking about the future around product analytics, is that this snowball effect will result in a situation where we can’t ever realise the value of event tracking.

[00:39:28] Bhav: So I’d love to hear your thoughts, Timo on like, how do we, how do we account for this? And let’s face it, like, I know you’re probably going to mention documentation, but documentation is one of those like things that never works. And I’ve heard about documentation from engineering teams and data teams and whoever it is, but documentation is one of the hardest things to get right. So how do we solve this? 

[00:39:48] Timo: I will not mention documentation, I might mention it in the end. I think I’ll mention something else, which can solve the problem. And it’s also sounds similar, boring than documentation in the end is like design. I think design is the only way to solve it. And so design means it’s like, you have to have a proper approach, how to define these kinds of events and the topic that you mentioned is definitely something that is haunting me for the last four years, I would say so, because I was guilty myself. 

[00:40:18] Timo: So especially like when you work as a consultant I build event data designs for companies. So like, okay, what should you track? And because I wanted to be a good consultant, I say, okay, you pay some money. And so you should get the best possible, and I define the best as the most complete tracking design picture that you can ever paint it. And it ended up in like, in the bigger project, we ended up with 120 unique events and development hated us deeply. 

[00:40:46] Timo: And I think like one of the moments where I really then afterwards took a step back and thought, okay, I have to think about this. Was like, when I did one project with a company where we had 40, 50 events or so we thought it’s really good, it’s covering everything that they do in a product. It’s was a complicated B2B software as a service product. The development team was on board. They implemented everything perfectly. We use all the best practices, how to implement all this stuff and so on, but after six months, no one made any use of this data. 

[00:41:20] Timo: The data setup was really, really good, but no one was using it. And this was a point where I thought, okay, there’s something wrong. And I think my explanation to this is like the approach, how you define events. So the approach that I did in the past that most of the teams are doing is like they do it based on the application they have. So they go in the application, they look and they select, okay, these are important things that people can do. 

[00:41:41] Timo: So they can click here, they can do this and so on. And by that, they define the different kinds of events by just going through the applications and the problem is like, when you do this from an application side, it’s disconnected to the business because like in the end, no investor reports is ever built on who clicks where on a navigation, so no one actually cares about this and what I started to test out in different projects then was like to go from the business to really say, okay, how does the product work?

[00:42:11] Timo: And to really understand how does the product work and derive events from there. So it’s the same example like with the features. So in the end, pick something which is very close to the business and derive events from there. And just by this kind of exercise, you control how many events you will do. And then the final trick on this kind of the secret sauce, because like In any projects where there’s always this one person who comes up like, yeah, but I want to understand who clicks where in the navigation, this is so important for me.

[00:42:38] Timo: And you have to respect this, for this person it’s really important because for some reasons, maybe they work on the navigation at the moment, or maybe they’re responsible to decide what should go in there, have a catch all very easy structure that can capture interaction events. So I always introduce a schema, which is called event clicked, submitted, hovered, whatever. And then I put the context of the properties. So like, okay event click could be like, okay, type navigation, place navigation, target, wherever this link goes to text. 

[00:43:10] Timo: Whatever text was there, even put colour or some other stuff in there, you can dump as many stuff in there and if someone actually needs this, they can find it there because this kind of, let’s say, noisy data is often interesting for two or three weeks and afterwards, no one cares about it anymore. Maybe then it comes back six months again, when someone is working on this kind of stuff. So we have this catch all kind of bucket where you can put it in and you don’t destroy the other event schema of important stuff, then you can handle and scale this pretty easily with that. So this is my solution to this. It’s not the perfect one, but it definitely works better than the other. 

[00:43:48] Timo: Yeah and in the end, like, because you don’t have so much things. So if you have a good design, this is already part of the documentation. So then you design first and then it’s already there. So the documentation is automatically there. 

[00:44:02] Dan: Well, yeah. And also there’s an element of like rolling this out and training and education and upskilling internally, which is part of the design and getting people on board because documentation I feel is only done if you’re doing it in a silo and you need to show other people what you’ve done and you’ve not brought them on the journey, you’ve done something, you need to tell them what you’ve done.

[00:44:18] Dan: And actually that’s not, we talked about this earlier in this episode, but like, that’s not the right way. It’s about bringing people on board, understanding what they need and designing for that right. 

[00:44:27] Dan: All right, Timo this has been a wild ride of conversation that’s really, really interesting. And I feel like there’s so many other threads that I want to pull on, but I’m very cautious of the time we have left. So I’m just going to promise you now that we’ll do this again and I’m committed to the audio and video format. So I have to follow up on that. But just before we wrap up the episode with our standard rapid fire questions. This is just your opportunity. Where can people find you? What do you want to plug? What do you want to share? And how do people get in touch with you if they, if they can? 

[00:44:54] Timo: I think the the best way to get in touch is like LinkedIn because like it’s still the platform where I’m most active. So if you have any kind of questions, always feel free to directly message me. I still try to answer them. Maybe it can take a week, but, I still go through the messages and I will answer. Another thing just because like the design topic is so important for me, this is like why I write a book about event data design.

[00:45:16] Timo: And so you can find it on timodechau.com/book and I basically try to put everything, which I was thinking about the last four or five years, and it’s a work in progress book. So in the end, you can buy it already. Now you get, I think 250 pages already, and you get 20 to 30 new pages every two weeks. So it’s like it’s an ongoing journey, at least for this year. So at some point I might have to stop it, but there’s still some stuff that I want to cover like implementation part, monitoring part but the design part is already almost complete in there, so if you want to do that.

Rapid Fire/Outro

[00:45:53] Dan: Amazing, we’ll put links to these in the show notes for anyone wanting to find that out. So without further a do, Timo, let’s go into the rapid fire. Let’s put you on the hot seat and get your rapid responses. So the first question, what is the biggest challenge today that will be gone in five years time?

[00:46:12] Timo: I make a hot take, the amount of data engineering that we have to do today will be, it will not be gone, but it will be significantly reduced to a lot of things that are happening right now. So be it like AI makes it easier to consume APIs and so on and other things. So I think like the, the painful stuff that we still have to do in data engineering will become a lot less, okay, crossing fingers.

[00:46:38] Dan: That wishful thinking, but it’s all good. So in this future where there’s less engineering work required, what will be the biggest challenge in five years time? 

[00:46:46] Timo: I think in the end, what always is the biggest problem making decisions, making good decisions. I think decision making will not go away. I mean, of course, like we will have assistants that can prepare this decision. Maybe we have more automatic systems that would do some decisions for us, but the big decisions are still on us. So. 

[00:47:05] Dan: Yeah, that’s great. So what is one myth that you want to bust? 

[00:47:08] Timo: That server side GTM is actually useful. So it’s like, yeah, I think they are, they’re definitely edge cases where it is useful. Maybe they are a little bit bigger than edge cases, but. how people at the moment think that server side GTM is useful, it’s definitely not useful. Server side tracking, which some people mix up, by the way, is useful. Can be really useful, but people are sometimes, I think maybe they’re just mixing things up so.

[00:47:38] Dan: Well, everyone needs to get CAPI installed, right? That’s what Facebook’s telling everyone. 

[00:47:42] Timo: Yeah don’t mention CAPI so it’s really like, it’s yeah, it’s a trigger word, definitely. Especially how it’s often implemented. 

[00:47:51] Dan: Oh, for sure. Yeah and doesn’t, doesn’t improve anything. Anyway if you could wave a magic wand and make everyone know one thing, what would it be?

[00:48:03] Timo: And maybe this is including myself. Be better in taking a step back and look at the bigger picture where you’re like this, I think it’s so easy to get wind up in these let’s say immediate matters like, oh my God, I’m losing data. Oh, this is like, so problematic. It’s like taking the step back and like say, oh yeah, maybe we didn’t even need that one so it’s totally fine. I need this as well from time to time, so sometimes to focus on stuff. 

[00:48:32] Dan: For sure, I just want to add a bit of context from recently lived experience for me is that when we get people ask us about exporting all their old Universal Analytics data from like 10 years ago, and they’ve never once needed it before, but they think they need it now and just take a step back and realise that you didn’t need it before, you’re not going to need it in the future. 

[00:48:47] Timo: But in general, year over year numbers often doesn’t make sense. So it’s really like, it’s rare that this kind of report helped anyone, I don’t know. 

[00:48:57] Bhav: Especially now I think we live in a world where year over year definitely doesn’t make sense. I think there was an argument maybe pre pandemic, but yeah, I think that’s a time long gone now.

[00:49:07] Timo: If you want to get a budget raise, maybe then you can manipulate some year over year numbers and make a case for you. But yeah this is a different topic. 

[00:49:16] Dan: Yeah, you need those historic data points to do the projections forward, right? That’s how, that’s how it works. Anyway, Timo, last question and probably the hardest one, and then you’re off the hook. What is your favourite way to wind down outside of work? 

[00:49:26] Timo: Oh, good one. I don’t do this so much, but usually like it sounds ridiculously I don’t know, weird, but gardening work is definitely good to do that. I don’t have a garden at the moment, which is definitely a problem. But I should get one again. So, but this is definitely something, if you really want to get me totally away from even thinking about any kind of, okay, how could I do this? This is the best way to do it so. 

[00:49:51] Dan: Amazing. Well, I’ll kind of join you on that. I don’t have a garden, but I really love gardening. I love the idea of it and I definitely see that in the future as something to kind of like, oh, I’m going to be that old guy with a garden that’s like every single day being touched. 

[00:50:02] Timo: I had this in the past. So we had some time where, and this was always the best. So it’s like, it’s immediately shutting off everything. So I should get this back yeah.[00:50:11] Dan: Amazing, you’re off the hook Timo, thank you so much for this amazing conversation. And for everyone listening thanks for making it to the end, everything’s in the show notes. You’ll see the link in the description of either the video or whatever podcast app you’re listening to. Yeah, again, thank you so much Timo.

Tags: BigQuery data analysis event tracking Podcast product analytics product management The Measure Pod
Share:
Written by

Subscribe to our newsletter: