Why and how we’ve chosen Superset as a general BI tool

The common question is selecting the BI tool most suitable for the company’s goals. In this article, I want to share Dodo’s approach to managing it and give examples of what might be significant in your team.

For a brief, I have to reveal how our business model works, and it will clarify some specific requirements for BI Tool we had.

Dodo Brands is developing by franchise model and operates in 17 countries. We have a great chain of partners to whom we provide our business model and technologies. Headquarters has been developing its own information system since the beginning of the company’s history. This system helps automate all store processes, create orders via mobile app, and observe every delivery. A business intelligence system should be integrated into every function of our company to allow look for new insights and help improve our service.

The company’s business model and approach make us look for the software that should solve the problem of creating new analytics and providing it to thousands of our colleagues. Of course, we had a few solutions, and let me explain them.

What we had and how it looked like

As I said before, we’ve been developing our information system, and it has a part called “Reports”. Our software engineers developed this part, and it looks like a general tool to load reports of prepared SQL queries and developed visualization that you couldn’t change. To change anything in this part of our system, the task should be created and prioritized in the common backlog. However, only significant changes could appear in this backlog and, after that, be taken into work, developed, and deployed to the production system.

What does this approach lead to? There is a significant gap between creating new functionality and when our partner can get any new analytics. For example, we had one year and a half between launching the function of selling “Combo Products” through our mobile app and creating reports for our partners about revenue from these products.

I’d be lying if I said that we didn’t develop analyses. We had the BI department and used Power BI inside headquarters. However, we didn’t have an opportunity to share access with all of our partner’s teammates because it cost a fortune. And we couldn’t share these costs via Royalty or pay for licenses from the general budget. Could you imagine the final price if you have 20 thousand employees in your chain?

The only solution we found offered us to buy licenses for each employee. And only AWS allowed to pay for the session at the moment, and you don’t have to pay for every personal access. However, we are working on Azure Cloud and didn’t have plans to complicate our infrastructure with a multi-cloud approach.

All vendors we talked to and asked to change their monetization approach rejected it. I understood that, like BI tools, providers have several enterprise customers who pay them the most of their revenue and are not interested in developing a market for small businesses. It’s not my place to judge big companies’ strategies, and I wouldn’t, anyway.

The decision we made

The payments for BI tools licenses are only the tip of the iceberg. We looked for solutions to solve end-to-end data-based problems. Our team of excellent specialists worked on building an infrastructure that provides tools for creating a distributed system for Machine Learning, Dashboarding, and Ad hoc analytics solutions. We had to find the most suitable for our business model approach.

I believe we did it. Later I’ll tell you about all parts of our Data Platform. Meanwhile, let’s come back to the BI tool part. We decided to use open-source solutions. However, do not make a mistake. Systems that are open for contribution and free use are not cheap for your company. It will be paid in different ways: employees’ salaries, time for developing necessary functions, maintaining the system in production, and creating supporting processes. Sometimes, it could be more expensive than paying for licenses. Added all cons and pros, we decided to make a BI part of our Data Platform based on an open-source solution and started looking for the most suitable for us.

How to choose an open-source BI tool and how we did it

After some research, we filtered two open-source solutions: Redash and Superset. These were our primary selection criteria, which you can take as a basis for creating your own:

Visualization tools differences:

• Redash is more straightforward and flexible. However, it’s impossible to create something without SQL skills. The interface is user-friendly, but the visualization tools don’t have all the necessary charts;
• Superset is more complicated because of various settings and semantic layers. However, it gives opportunities for some simple calculations and analytics without SQL knowledge.

Summary

Our examples revealed the reasons to use open-source projects and contribute to them. We integrated Superset into our Dodo Information System and customized it for our necessities. This solution allowed us to avoid any vendor lock in the future of BI tools in our company depends on our plans and roadmap.

Integrating SaaS products into a company’s processes and paying for licenses could be the right choice with fast realization. However, suppose you have a clear understanding of financial advantages and how the flexibility of open-source could be used in your business model. In that case, it might be a great idea to start contributing to open projects.