As a Product Leader, knowledge is key.

You need to understand how customers are using your product. What their journeys look like. How you fit in that journey. Why they behave the way they do. What keeps them engaged, Or what’s pushing them out.

That kind of knowledge informs product strategy.

For practitioners of Lean (Startup), that’s in line with the following core principle:

The unit of progress for Lean Startups is validated learning

LeanStartup principles

To gain that knowledge, you may develop close relationships with your users. But when you’re starting to deal with more than 10 users, you need a more robust approach to learning.


Learn Fast

Maybe Google Analytics is all you need at this point.

I mean, why not. It does provide a ton of value with a very simple setup. If you’re able to generate enough knowledge through Google Analytics to keep you moving forward, I think you should stick to it and squeeze that lemon.

Because, keep in mind that the main focus of your analytical journey is to gain knowledge quickly and iteratively. If Google Analytics does the job, that’s awesome!

But there is a world beyond google analytics.

Google Analytics provides an aggregated view of your user’s behaviours. However, you eventually may want to understand how each user behaves. How does their relationship with your product evolves. There’s a whole new level of learning to be gained by having access to such granular data.

And this is where you should start thinking about going beyond Google Analytics and develop your own analytical stack.


Architecture Of A Product’s Analytical Stack

Your customer data is usually dispersed in many data sources. The challenge of piecing them together can be quite complexe.

Artwork by Micaela Lattanzio

(Artwork by Micaela Lattanzio)

So how do you get from the individual pieces of data to a cohesive view of a user’s profile and behaviours?

Customer Analytics Program - Architecture

Customer Analytics Program Architecture from past project

Above is an example of stack we created for a project. The goal was to create a customer analytics program that could let product leaders consult how users were interacting with their product through multiple touchpoints. As you can see, the infrastructure provides a lot of flexibility to integrate new sources of data, as well as pipe clean user data into multiple BI tools.

Now most of those layers actually requires very little work. Segment and SaaS BI Apps can provide on their own a very quick way to get to a very rich view of your user’s profiles and behaviours.

But if you want to get into more custom analytics, you’ll probably want to build your own data warehouse which will hold clean and massaged data that can then be consumed by your own set of BI reports built in Tableau, Looker or Mode for example.

And we’re not even talking about other possibilities having your own data warehouse opens up. If you ever want to get into data science (experiments, predictive analytics, etc.) and maybe even AI, then you’ll need a clean repository of your product’s data on which to run experiments, train models, etc..

So, at one point in your product analytics journey, learning will require more sophisticated means, and that’s where you’ll need to invest time and development effort to create a data warehouse and reporting suite that fits your needs.

Luckily, again, cloud services and open source projects abounds that lets you do this fairly easily.


Modularity And Scalability

As you grow your analytical stack, it’s important to make choices that will be scalable for you future needs. Introducing modularity in your architecture is an important consideration.

For example, in our example architecture above, we could (somewhat easily) change the Segment module with something similar such as Snowplow. We could interchange data warehouses, with cloud-based solutions such as Redshift, Google BigQuery, Snowflake, or just a really simple Postgresql database.. Even your ETL pipelines could be built using your own Python scripts (beurk), or cool new transformation tools such as dbt (yeah).

The point is that you don’t want to be boxed into an enterprise-level solution.

What would be the point anyway when there are so many tools available nowadays that allows you to build a scalable and cheap architecture quickly.

And if you don’t like how one module works? Change it!


Getting To Know Your Users

Artwork by Charis Tsevis

(Artwork by Charis Tsevis)

In the end, no matter the services you choose to build your modular, cloud-based, modern, scalable and cheap analytical stack, the point is always the same:

Gain knowledge about your product’s users, quickly and iteratively.

So now that you have your analytical stack in place, time to leverage that data to better understand your users.

It’s only when you start piecing customer data together that you get a sense of their individual stories.

Mapping the journeys of your customers; segmenting them by demographics, behaviours, psychometrics; identifying their interests through what they read on your product’s blog, or what they tweet about, etc., etc., the possibilities are endless.


Orchestrating Journeys

And finally, remember, the goal is not just to be amazed by the data you now have. It’s to act on the insights you have at your disposal.

The art of acting when needed – to the point of automating personalization with such a tool as Segment Personas or Snowplow React.

Better understanding your product’s users should have for purpose to engage intelligently and help them move forward on their journey.

It’s always a game of providing value quickly and iteratively.

What do you think your users want? Put out an experiment that tests your hypothesis. Look at your analytics and see if there was an impact on all your customers and just that segment that received that new treatment.





On-Demand Product Analytics Expertise

Lantrns Analytics provides consulting and development services for your online product’s analytical stack.

All product leaders understand the value of data-driven decision making in regards to their product’s development strategy. But building analytical stacks for your product requires experience and knowledge that may not be readily available within a team. You can sometimes get away with cookie-cutter solutions, but should be mindful that although your analytical stack aims to fulfill current needs, it should also scale in function of future needs.

Lantrns Analytics provides on-demand experienced staff to supplement your team’s needs all along your analytical stack.

We help you build SaaS-based, modular BI infrastructures that fits your current needs and is scalable.

We’re in it for the long run, from strategic consulting, to development, continuous improvement and support of your product’s analytics stack.

Book a time to chat with us