Getting the most out of your data: setting up a BI Stack 101

Many app developers can find themselves constrained by their data management systems, especially as they grow beyond set-ups that work for smaller apps. But, if you are a firm that has decided to level up in terms of data analysis, that poses the next question. How do you set-up a BI stack that works for your app?

A BI stack is all the different layers of software and databases that make up your company’s ability to ingest and process business data. Author and co-founder of data consultancy Tale About Data Lior Barak spoke to Adjust, explaining how firms can implement a BI system that is suitable for your use case.

“In the past, a lot of IT projects started with the question: buy or build?” Lior explained. “Both of those options meant making a big up-front investment.”

However, developers shouldn’t be scared off by the thought of hefty prices. Nowadays, Lior explains, the best choice for most companies is to leverage the Cloud. Developers can get lots of bang for their buck by utilizing well-established software ‘building blocks.’

Building blocks are free or paid scalable components provided by your cloud provider that allow you to store, process and display data in an easy way.

“Many of them are on rented servers with a pay-as-you-go model,” Lior added. “This way your capacity can be seamlessly scaled up and down based on your needs — as you go from your first user to your 10 millionth.”

There are also big benefits to thinking deeply about your BI stack from the very beginning. A lot of companies originally set out to build a decent data infrastructure and soon find themselves trapped reinventing the wheel and managing infrastructure — even though none of that was in their business plan. So how do you make sure your focus is on your app — not getting lost in data management?

While you don’t want to lose track of your actual mission, you also don’t want to regret not having collected historical data from Day 1. So what are the essentials? Lior says there are three things to keep in mind when starting out.

1. Store all of your data in one place and set up proper tracking: The best data warehouse in the world is worth nothing if you don’t fill it with accurate data. Remember: if you don’t collect your data today, you can’t analyze it tomorrow.
2. Privacy first thinking: When you design your data, think about the user's privacy. How do you make him feel comfortable with you sharing his data, and how do you secure the data transformation as well as the data storing. Data with a privacy-first approach will raise limitations, but also create new opportunities.
3. Connect every data source: Too much time and human potential has been wasted on copying numbers from one Excel sheet to another. By connecting multiple sources to your Data Warehouse (for example, via APIs), you can create a single source of information for your organization.

So once you have set up our single source of truth, and accurate data is starting to flow in, how do you turn this into a proper reporting landscape?

Think about aggregations: If you aggregate your data on a user-level, you can look at numbers like Lifetime Value or Retention, while at the same time your finance department can see their numbers aggregated on a monthly level to see how much money is flowing into and out of the organization. That’s what we call data modeling — the process of creating and defining systems or structures in a database to support your business goals.
Keep your raw data: Once data has been ‘flattened’ by day or month it can’t go back, so keep a record on the most granular level, because cloud storage is cheap and you can always aggregate up later.

With these tips in mind you can ensure that as soon as you start collecting data, you are set-up for success.

You can find a complete list of analytics platforms (BIs) that are supported by Adjust. If your selected BI system has a module integration set up, you can find a step-by-step walkthrough at https://help.adjust.com/en.