Microsoft is definitely on the right track. Thanks for sharing your insights Teo!

It would have been even greater if Microsoft provided a 64 bit client with pivot table functionality for pure self-service, like Tableau does. 64 bit Office is not something that most organizations will install, and server-based solutions require more from IT. The Large Address Aware capability change for Excel 2016, which gives us access to 4GB virtual memory for Excel 2016 32 bit Click-to-Run versions on 64 bit Windows, will be a great improvement though.
https://support.microsoft.com/en-gb/kb/3160741

That’s where the forthcoming upgrade to SSRS comes in as it will allow us to publish PBIX files to SSRS in native mode.
https://blogs.technet.microsoft.com/dataplatforminsider/2015/10/29/microsoft-business-intelligence-our-reporting-roadmap/

It looks promising! I noticed that it’s possible to create calculated members (as well as calculated measures) using MDX. Some people will surely appreciate this. 🙂

Power BI introduced Analyze in Excel that lets business users create pivot reports connected to published datasets. Have you tried this feature?

https://ideas.powerbi.com/forums/265200-power-bi-ideas/suggestions/6881220-integrate-powerbi-with-pivot-tables

Tableau is a decent product when it comes to creating dynamic crosstab reports, while Microsoft unfortunately has nothing better to offer than matrix reports.

Thanks for the feedback, George! About 2, can you describe a scenario that requires “post calculation”? About 3, could you point me out to the “wrapper application”. I found the Tableau REST APIs to be similar to the Power BI REST APIs.

As for Tableau, I fall in love with it with the first 3 day’s trial two years ago. I enjoy playing with Tableau and you never feel boring with it. Personally, I think Tableau’s modelling capability is far behind SSAS/Power Pivot. You can only have a single fact table based star schema. The LOD expression is powerful but only as half powerful as MDX/DAX can achieve. I don’t really think Tableau is multi-dimensional like true OLAP engine. It is a 2D tabular data with fake OLAP appearance. However, the UI is very flexible, it renders quick and beautifully. You can place UI controls anywhere on the canvas, you can change it in different layout, e.g. drop down, sliders, check box etc. The data point selection is awesome, I can use the lasso control to group items and create groups / sets on the fly. I can then use the groups / sets immediately and create hierarchies on the fly. The sets can perform union, inner join, exclude with user interface easily. The chart supports nested in rows and columns and therefore the trellis chart type of Vis made easier. The parameter can be used everywhere, such as data source filtering, calculated fields, reference lines, table calculations etc etc. I also like the custom shapes. Tableau does not have native gauge chart. I use SSRS data driven subscription to generate 100 images and use the images in a dashboard.

Three things I think Tableau is good but Power BI is lacking are:

1. The “Web Page” component in the dashboard. This allows Tableau dashboard can embed other BI assets such as SQL Server Reporting Services reports, SAP Business Objects, R / Shiny / Plotly, Google Charts, D3, virtually any web reachable content, with dynamic URL controlled by the Tableau expression language

2. Table calculation. MDX / DAX is powerful, actually much more powerful than Tableau’s expression language. However, it is single layered. The calculation is performed in the storage engine and formula engine and returned to the client. That’s it. In Tableau, the table calculation is some kind of “post calculation”. The regular calculation is executed and return an intermediate tabular structure and then table calculation perform another calculation against this intermediately layer. I understand in Power BI, the R integration is similar to the table calculation which is a post calculation. But in power BI, we don’t’ have this “post calculation DAX” yet. Maybe the newly introduced “calculated table” can achieve this but I have not tried. And, it is a lot of DAX code to mimic this post calculation as you have to deal with the row/evaluation context and relationships.

3. Tableau has rich JavaScript API to allow embedding and controlled by the wrapper application. I can easily embed Tableau in custom ASP.NET, SharePoint Online, PowerPoint link: https://george-qiao.com/2016/04/29/embed-tableau-into-microsoft-power-point-2016/) and hopefully Office Sway soon.

People keep blaming Tableau has very poor data preparation capabilities. I agree. But I have strong SQL Server data development skills and I can leverage T-SQL and SSIS to achieve anything I need. I just use Tableau for the “last mile”. I wish one day, Power Query can be published onto PowerBI.com and exposed as an ODATA endpoint. Then I can use Tableau to connect to an end user self-prepared data pipeline.

That is just my personal experience. I love Tableau’s flexibility and simplicity. By no means Tableau is more powerful than Power BI stack. Tableau to me is a great UI tool to help me to explore the data and get the results quickly.

Thanks for the feedback, David! About Power BI maturity, I tend to disagree that Power BI is lagging. True, Power BI Service was introduced less than an year ago but its building blocks (Power Pivot, Power Query, Power View, Tabular, Azure) have been around for years.

I’d love to get your input about other areas where Tableau surpasses Power BI. It appears that Tableau considers these 10 points to be the most significant to prove that Power BI falls short and that’s what I reviewed.

I would agree with your assertion that some of the points that Tableau makes are insignificant. For others though, Power BI lags behind primarily due to the immaturity of the product rather than a lack of vision. I doubt anyone at Microsoft sat in a planning meeting and said that they don’t ever want what-if capability or more convenient, auto-generated calculations. Feature gaps can be filled quickly if needed and will not offer as much differentiation as Power BI matures.

Of course, most of the items on the list of why Power BI currently falls short could have been applied to Tableau itself in its early years. What-if scenarios? Tableau did not introduce parameters until v6 after six or seven years of development. Analytics? Most of the great statistics and forecasting came in v8 or 8.1 (as an aside, I think you have understated Tableau’s capability here as it offer much more than trendlines and forecasting. It would be better to compare it to the Analysis Toolpak in Excel rather than anything native or coming soon to PBI). On the other hand, Tableau did not add the ability to *pivot* data until v9, didn’t have *filled maps* until v7, and spent much of 2015 celebrating LOD expressions, etc. It will be interesting to see how a list like this might change over the next year or two.