How to chose: Tableau vs Power BI for enterprise decision making
Salesforce’s Tableau and Microsoft’s Power BI are both business intelligence tools that help with visualizing patterns in data through a GUI drag-and-drop interface. The goal here is not to pit one against the other, but to provide some reasoning to consider in the backdrop of each unique enterprise.
The key difference is that Tableau is a great story telling tool with a focus on easy visualization, where as Power BI gives more power in the hands of a data analyst justifying its eponymous name.
This simply means there are different skill requirements for each tool and that impacts why we might choose one over the other.
Skillset
Tableau’s focus is on ease of visualization. There are very little pre-processing, data modeling or data engineering options. If the enterprise doesn’t have experts in data modeling, SQL and ETL, installing Tableau into that environment might create some technical debt.
That’s because Tableau expects the data to be provided to it in a format (usually a denormalized flat structure) that’s easy to visualize. The tool is not super comfortable with having to do joins or data modeling. Only recently did Tableau (close to 2020) made changes to its data join features to be more robust.
Power BI provides data modeling and pre-processing features. Neither features are close enough to be used as enterprise-grade data modeling or ETL tools. It sill gives enough tools for a strong Data Analyst or for a Data Engineer to wrangle data before visualization.
In a nutshell the ideal target user for these tools have different skillsets.
Intuitiveness
This is a double-edged sword based on which side of the skillset you’re on.
For a business person or an analyst that’s not tech-savvy, Tableau might seem intuitive. It doesn’t bog you down with what I call “thinking in sets” mindset that’s needed for SQL expertise. With Tableau we don’t need to bother too much about sets of rows, aggregations, cardinality etc. Simply click or drag a measure or dimension, and Tableau arranges and aggregates those data fields intuitively.
But for a data engineer that intuitiveness might be a bit strange and hard to digest, at least initially, because “intuitive” usually means hiding complexity — which hides some detail that a data engineer would appreciate.
Power BI on the other hand let’s us think in terms of data models and sets of rows. We create reports as if we have just produced an SQL result set which needs some visualizations — it’s a natural progression from result set to visuals, at least for a data engineer who has been dealing with row/column result sets.
Excel aficionados
It has been discussed time and again that Microsoft Excel is the ultimate data science tool, without the sex appeal. There are so many blogs out there arguing that Excel is capable of doing a lot of what sklearn does, at least for most enterprise data analyses.
Data Analysts migrating from Excel-based analysis to a BI tool might greatly benefit using Power BI due to the unified user interface across Microsoft’s productivity tools. Power BI also extends some of Excel’s advanced data manipulation features.
Other considerations
Isn’t it better to invest in a BI tool that does one thing really well — the only thing it’s supposed to do — which is visualization? Probably true. And that tool is Tableau. If there is a strong data engineering team available it appears all the more true since the analysts can just focus on visualizations with ease.
There are at least three arguments against it.
One is evident — cost. Tableau is expensive. And like many other Microsoft tools, Power BI isn’t. It’s free to download. Only publishing the work to other users requires a license.
The second argument is that both tools are largely similar. For common enterprise requirements, both tools provide decent dashboards. Both tools run faster when data is being read from a local native file format (.twbx, .tde, .hyper for Tableau and .pbix for Power BI) as opposed to reading from a database. Both Salesforce and Microsoft are competitive in adding advanced analytics and machine learning features to the tools.
The third argument is again skillset. As a startup you probably want your 2 business analysts (or the 2 people that does everything in a startup :) ) to be able to use an easy visualization tool — Tableau. It’s great if these 2 individuals stay with you as long as you want. When you hire new people to build dashboards and reports, they come from crude technical background — regardless of the type of BI tool you hire for. So “easy visualization for business users” argument goes away.
On the other hand if the investment in Power BI is because there is no data engineering team — perhaps the goal is for one engineer to do both data wrangling and visualization using Power BI — there might be bigger problems lurking. Lack of a strong data engineering team is recipe for chaos.
Say no to fringe features that won’t be used today or in 1 year
I see all BI tools as SQL visualization tools. They add color, shape and geometry to group-by’s, aggregate and analytic functions in SQL, making it easy and intuitive for people to visualize patterns and arrive at conclusions about data — without worrying about SQL.
As with all of computer programming tools, BI tools have universally made the promise of enabling business users to be creators. This is why some BI tools tout features like “semantic” model — an interface to convert data fields into canonical names for business users. However, as always, these features end up being just another thing that tech people have to build and maintain for no realizable value.
Looker seems to be at a point where Tableau was 10–12 years ago. A lot of not-too-big-to-fail companies on strong growth trajectory are using Looker.
In the Gartner report above, Tableau and Power BI lead the way. It would be interesting to see how Looker catches up. Acquisition by Google might be Looker’s boon-and-bane dichotomy— the investment is great but being part of the #3 cloud provider may thwart its momentum.
