There are quite a few articles and blog posts about Power BI vs. Tableau, and I have no intention of writing another one that will compare features, prices, ease of use, learning curve, and adoption rate.
Most comparisons reach the conclusion that while Power BI isn’t as rich yet (They all say “yet” and they are all correct), it’s growing fast, and for a price tag of 10$ a month per user, it’s the winning horse at the moment.
I’ve worked with both and I think they are both amazing. Power BI is my current go-to BI product because it’s just way easier to approve a small budget, and that’s what Microsoft is thinking. They give you a 60 days pro trial and then the monthly subscription can simply be stopped.
But it is limited(still) and there are some extremely important missing features(the ability to save filters, for instance), so while I’m happy with Power BI, I’m not married to it and I’ll be ready to ditch it when there’s something better with a similar price tag.
So while Power BI is becoming more and more popular, and while my tendency is to offer Power BI as a solution, a different angle to this debate is to look at the history of BI visualisation products and learn an important lesson. I haven’t used all products but I’ve used many of them and it seems like every 3-4 years there’s a new king or a new debate regarding who’s better.
It was Oracle Discoverer 20 years ago, and Business Objects, and SSRS, Cognos, OBIEE, Qlik, Tableau, Power BI and I’ve left out probably 250 others.
There will be new products and old products will get a fresh look, and this industry will keep evolving.
If you need some of the features that Tableau is offering and Power BI isn’t, and cannot do without them, buy Tableau. If you can do without, buy Power BI.
It doesn’t matter that much.
What does matter is that you prepare your underlying data the best way possible. Then you can plug whatever you want on top of it. Tableau, Qlik, Power BI, Excel, some proprietary code in C# just because you can and maybe expose your data to 3rd parties if you need to via APIs.
I’m not impressed with the modern tools’ modelling features. They all offer the ability to load data from disparate sources, clean and blend everything and then quickly analyse. Let me refrase. It’s not that I’m not impressed, I hate it.
If you use Power BI or Tableau to load data from excel, mix it with data from a database, maybe throw in some online data source for good measure, and then create 25 calculated columns on top, that’s just the wrong way to prepare data.
You’ll end up repeating the process again and again for different datasets and reports, and you’ll be 100% dependent on the product that’s currently the king of the industry.
Instead, I believe it’s better to do the modelling in the database. An old-school Data Warehouse, or modern, smaller data marts which are built in an agile fashion can be reused by any report and any reporting tool.
Here’s a recent example. A client of mine asked me to prepare a sales report for a specific department in the organisation. Sources for the report included 2 excel files with some metadata and one operational sales SQL database. They use Power BI, and I could have created a Power BI dataset that reads from these three sources, add a few calculated columns and measure, then create the report.
Instead, I used T-SQL to load all data into a data mart and prepared all the calculations in the database. Then, the Power BI work was pretty simple- just loaded the ready-made data mart(a star schema with 5 tables) and created the report in a matter of minutes or hours. A week has past, and they asked for a similar report for a different department. The sales source database is the same, but the metadata sits in other excel files with a slightly different structure. I added a few pieces of code to the ETL T-SQL process and added the new data to the “old” data mart. Then the Power BI report needed zero changes to support the new department. It just worked.
Had I done everything in Power BI, I would have had to repeat most of it for the second department, and then maintain two versions going forward.
If you are like me and you understand the importance of investing in your data warehouse/marts, you’ll be able to use whatever BI product suits you, and you’ll be able to switch between tools if needs be. It doesn’t really matter, as long as your infrastructure is ready.
Additional reading:
