When you should consider migration to open-source BI tool

Last week I participated in an amazing event organised by Aviel Ilan. One of many things we discussed was data literacy and how to encourage people in the company to use data in their daily work. Filip Vítek brought up one thought which I personally found worth sharing. It is about open-source free BI options and their benefits.

To me, usage of free open-source tools is always a trade-off of the price you pay to a vendor for a managed service, availability of technical support, and usually a better feature set. And the cost of additional resources you have to have in order to support open-source tools by your team keeping in mind missing opportunities with your team to work on maintenance instead of adding more business value.

Filip’s point was with open-source BI tools you can easily give access to the data to more employees at no extra cost. And hopefully, get extra business value from allowing people to make data-informed decisions, again, at no extra cost. In many cases, it may compensate for the scarcity of features and lack of support.

To me, it was a clear but underestimated benefit. Usually, when you start BI tool implementation the first step is to figure out how many licences of each type you need. In many cases, you have to limit access to the system to keep the cost reasonable. Then, when new people come, you have to review the access periodically and buy more licences when there is no other way.

But what if you can give access to everyone with no struggle and with no cost?

We can roughly estimate at what number of licences it is more beneficial to go for open source tool. For easiness of calculation let’s imagine we already had 100 users in the company.

I’ll use Tableau as an example of a paid BI tool because I’m familiar with their pricing policy more than with others. The current price of the Tableau Cloud licences is 75 Euro for the Developer, 42 Euro for the Explorer and 15 Euro for the Viewer per month. The reasonable distribution may be 10/40/50, which gives us 3180 Euros per month for 100 users. For that money, you’ll get Tableau as a service which means no extra cost for infrastructure and maintenance. Also probably you would be able to negotiate the price, but let's put it aside for simplicity.

On the other hand, free options like Superset require you to set up your own server and the cost of maintenance is also on you. I assume 2 days of maintenance per month on average should be more or less enough, which gives us around 2000 Euros. Plus the cost of infrastructure. Let’s use 16vCPU and 128GB RAM, which costs around 1000 Euros per month. Around 3000 Euros in total.

So, for 100 users we get a more or less similar monthly cost.

If you need to buy more extra licences, probably it would make sense to consider an open-source option. But also keep in mind the cost of migration of the existing dashboard. A company with 100 users probably have around 50-100 of them. I really doubt that they can be migrated to a new BI tool for less than 3-4 man-months, which is another at least 20K+ and a really painful migration project when you have to support and develop existing dashboards and run the project at the same time. Never an easy job.

So what would be a solution?

There is no right or wrong one. It always depends on your situation, your budget, your number of dashboards and the data maturity of your company. My approach would be if I need a BI tool for less than let’s say 100 users and I don't expect that number to grow significantly in the near future, I’d go for a paid fully-managed service (of course if there are no other limitations). If more, and the growth is around the corner, then I’d add an open-source one to my TCO calculations together with the cost of the migration project.

The number of users can be lower or higher depending on data maturity and how many people can and want to create their own reports. The more mature the company, the lower threshold would be.

All these calculations are pretty rough. Please don't use them directly. Always carefully use your own numbers. And good luck with your very own and exciting BI journey.