Meet the analytics stack

In this episode we have a discussion about the value of a data team in a company, which, naturally, is a huge topic. We talked about how new categories of analytics tools, like Kausa, are changing the role of data teams. Instead of just receiving requests to investigate issues, data teams can now proactively reach out to other teams with interesting patterns and work together to find solutions. This makes the data team a more proactive player within the company and changes the value perception. Overall, it was an interesting conversation about the potential of working with data in new ways.

Our major goal is to focus on our customers. I guess this kind of sentence you will find in any presentation of digital product vendors, maybe even beyond. So people always claim that they take the customers first, that they do everything for their customers. And yes, to some degree they do. In the end, they bring the revenue, and so they might listen to what customer says.But customer focus comes on different levels and talking to Maciej from Piwik Pro shows what real customer dedication looks like. He explains how Piwik Pro is really building around their customer feedback, and for their customer’s use cases. Everything is measured if it solves the problem or not. And you will see this as a constant pattern in this episode. So Maciej started out contributing to Piwik as an open-source project because he needed it for his own product. And then he started to build a consulting business around it. And at some point recognized that it needs a different product beyond the open source one to really solve the client’s use cases. And so it became Piwik Pro and got a completely new foundation. And from the early days, it was built for use cases where data is more sensitive as if you just buy shampoo. Like as government services, health care, or intranets. All before GDPR came around.And now Piwik Pro is on a clear growth trajectory with a lot of early investments and decisions paying off. I hope you enjoy the conversation in the same way I did.

There is a specific kind of person or a particular kind of role, which is usually ignored in data projects and especially in tracking projects. And these are developers. It's pretty shocking - I know. They rarely play an important role in these initial data projects. Even when they finally have an essential role. They ensure that the correct data gets sent to whatever tool you use.It’s often: “Take these 100 events and please implement them fast because we all need them for things you don’t need to care about”.And I can already tell you this is usually an approach that doesn't work out well. And that was interesting when I talked to James from Posthog and asked him, what is your target audience?He immediately answered: developers. We are a developer-first analytics product. This is quite interesting. Because product analytics targets product managers, usually. But Posthog calls itself a Product OS and takes the features beyond classic product analytics. So they are targeting the developers. Well, the persons who are basically building the product.And it makes sense. Developers are the ones who know exactly how a product works. They know precisely where it is the best place to maybe get some insights from. And usually, they have often really good ideas about data. What kind of event could be important to make the product better?This episode was a good lesson for me on how a product is defined when you have a clear idea of your target audience and how your product automatically starts to be different from the usual competition, just because you have a different focus. I hope you enjoy it as well.

Alexander from Elbwalker and I are talking about why tracking implementation is so hard and creates so much frustration on the way, why pure auto-tracking is not the solution, and why context is pure analytics magic.New tracking setups have a lot of really cool high-energy moments just because it's something new. And often we do a new tracking setup replacing an existing tracking setup that was not so good. And so everyone is super motivated in this to create something new and then comes the implementation.And implementation usually sucks up any joy and fun of a tracking setup projectBecause it's hard to get some resources in the first place and then it takes longer than everyone thought. The developers don't really like to do it. because for them it’s something they don’t do every day. And then things are not implemented in the way that you want to have them implemented. So for me, tracking implementation is still one of the hardest things to do. And Alexander from Elbwalker was so frustrated by this that he started a company to make it better.

Would you start a new product when the category is pretty crowded already What if you start a new product where the category is pretty crowded already?And most of the people that you talk to tell you: maybe not a good idea.But Oliver and Hamzah decided to do it. A new BI tool based on modern data principles and tools.It’s built on top of the metadata that you already have when you use dbt. In the same place where you describe your model with descriptions and columns, you can use Lightdash to define your metrics and dimensions and how these different kinds of data tables relate to each other.This is the metric or semantic layer everyone is talking about. An abstraction layer between your tables in your data warehouse and the place the business users are working with it.So marketing, product, or sales. They basically get a prepared version of the data just for them. With metrics and dimensions for them to work with. And so they don't have to think about how do I get the revenue and where do I get the country information so I can drill it down by country.

Listen to my conversation with Jonas from Weld to learn more about deciding on a data stack, the benefits of an integrated platform, and how to start with data in your company.It’s so much fun to build your data stack bundle by combining 2, then 5, then ten different tools. The significant promise of best-of-breed is that you shop for the best solutions and just connect them with the promise that you can always swap solutions when something better comes up. That’s the theory.So why would you use a platform that is doing 4-5 things that single tools could do for you as well?- You definitely save time by getting everything running- You don’t have to learn four tools, just one- You just make one contract/subscription instead of 4These can be compelling reasons. But in my conversation with Jonas from Weld, I learned about a significant benefit that is not obvious in the first place. The meta layer.What is the meta layer? An integrated platform knows everything from ingesting, modeling, storing, and activating. Because it all happens in one place, they can run monitoring on top and do deep lineage. They simply know how data moves from a to b to c. They can even know how much costs you generate at the moment in your cloud setup and can advise you on savings.This is an immense benefit over best-of-breed solutions where you have to create the meta layer yourself (or believe that data ops, monitoring, or catalog tools do it for you).

There is one trap in analytics - putting tools into categoriesAnd I am guilty here too.So we have tools for marketing analytics (mostly GA). And we have tools for product analytics. And we have tools for app analytics. And surely some niche ones for content analytics, sales analytics,...I think you get the problem. We are complaining about silos because of specific teams and then we do the same mistake again with analytics tools.Is Amplitude a product analytics tool - of course it is but it's also a tool that supports growth teams, a full e-commerce setup, and now even marketing attribution.When one thing should become clear after this episode then it's about improving things with data. Small steps by small steps. That's the right category - improving business & user experience analytics.Talking to Adam and John is a great opportunity since both cover different experiences and bring them together nicely. Adam with his background in the "classic" analytics world and John is one of the best persons to talk about how product and data can work hand in hand.I hope you enjoy this episode and it motivates you to think about enhancing your analytics setup to improve your business and user experience analytics.

Great data persons are also great storytellers. And if they can present a narrative of how problems with data quality can lead to wrong product decisions and can slow down everything, you can be sure to listen carefully.Stefania from Avo is a perfect interview partner - she had built the data team at QuizUp. She was making the journey through all things data at light speed. Then starting her startup and learned that data quality issues could blur any good product initiative. And so finally co-founded Avo - a tool to ensure data quality.And it does not ensure data quality from the technical side but also from the conceptual by helping teams create and manage tracking plans.

What if you had created a product 10 years ago that perfectly fits into today’s world because it has all the ingredients we are looking for in data collection:- privacy first- data ownership & transparency- open sourceBut let’s correct one thing. We are not simply talking about data collection. We speak about generating high-quality behavioral data.And of course, we are speaking about: Snowplow.Yali Sassoon joined me in the first episode of the “Meet the analytics stack” podcast. We talked about the origins of Snowplow. It was created in a world where closed and inaccessible analytics systems like Google Analytics and Omniture (now Adobe Analytics) ruled the space.So Snowplow should be open and transparent from day 1. It started as an open-source project, and it still is until today. It focuses on generating high-quality behavioral data:- with data pipelines you own and control- data enrichments you own and control- well defined schemas, so you only receive the data you would expect.Snowplow had all the ingredients from day one that we are looking for today in a GDPR world where data becomes a core asset, and parts of the data stack become critical infrastructure.Generating and collecting event data is a critical part of a data stack, so using an open-source solution here has plenty of benefits instead of a closed system. And if you don’t want to manage it, Snowplow is offering their behavioral data platform where they collect your Snowplow instance in your cloud project/account (yes, you are reading the right thing, in yours, not their).Enjoy the first episode of the new podcast - Meet the analytics stack.For upcoming episodes, follow the podcast page on LinkedIn and, most importantly, subscribe to the show wherever you listen to podcasts.