More Thoughts On The New Power BI Features

My last post on the new Power BI features announced at the PASS BA Conference was not much more than a list of bullet points written during the keynote. Now that the conference is over and I’ve had a bit more time to reflect, I thought I would try to come to some conclusions about what they all mean. And, of course, indulge in some wild speculation too.


Microsoft’s new-found enthusiasm for working on multiple platforms is clear from two things: first, the announcement that the iOS Power BI app will be coming soon (though we should not forget that this was promised a long, long time ago and is very late), followed by native apps for other mobile platforms; and secondly the work that has gone into the HTML 5 version of Power View. Indeed, the latter was demonstrated using Google Chrome to underline the point. Of course this is the only commercially sane direction to take but it’s very welcome nonetheless.

Power View new features

The cool new Data Exploration features in Power View, which allow you to edit existing Power View reports – creating new graphs and tables and merging existing ones – are I think only going to be available in HTML 5 Power View running inside a Power BI site. The same goes for the new time series forecasting functionality. The demos also seemed to make a point of the touch-friendly interface. Now I can’t imagine that Excel Power View will move from Silverlight to HTML 5 any time soon (the Office team are notoriously conservative when it comes to big changes like this) so maybe we should assume that, going forward, the main focus will be the use of Power View inside a Power BI site to build reports and dashboards rather than Power View inside Excel on the desktop? Maybe the HTML 5 version of Power View will be what is used in the touch-optimised version of Office that is slated to appear this summer? Who knows. I liked what I saw though and these additions will go a long way towards boosting Power View’s credibility as a client tool.

Time series forecasting

It seemed like Microsoft had abandoned ‘data mining for the masses’ after SSAS data mining failed to take off, but clearly not. The time series forecasting functionality seems very easy to use (you can read more about it here and here) and I got the impression that other algorithms might be added soon – maybe Project Sage is relevant here? Another question to ask is whether ease-of-use was the real reason why ‘data mining for the masses’ failed to take off the first time around? It might have been part of the reason but another factor must surely be that business users don’t trust predictions when they don’t understand how they are made, while the data scientists and statisticians are already using other tools that give them a lot more control for forecasting.

RIP PerformancePoint?

The first reaction of several people at the BA Conference (me included) to the new dashboard and KPI creation features in Power BI sites was that this was the replacement for PerformancePoint. Personally I never liked PerformancePoint much and rarely used it, and it doesn’t seem that MS has had much enthusiasm for it in recent years. I don’t think we’ll ever see it in the cloud either. Killing it off would have the added benefit of removing a client tool from a stack that has a confusing number of client tool options right now. However I got the impression that the dashboard and KPI creation features in Power BI sites, as demoed, were fairly basic and they may not be much more than widgets that can be placed on the Power BI home screen with nothing to tie them together, but I don’t know enough to judge properly. I think it might be better to think of Power View as the place to build dashboards.

Integration with old-school SQL Server BI

The ability of Power BI sites to host SSRS reports, and for Power View to connect back to SSAS on-prem, is an important bridge between what most of Microsoft’s BI customers are actually using and the new world of cloud-based Power BI that Microsoft is promoting. For me this was the biggest announcement of all. Will these customers be interested in buying Power BI licences for their users if SSRS and on-prem SSAS is all they want to use though? I don’t know, but I assume that this will also enable mobile access to SSRS and SSAS via the Power BI mobile apps, so that will be a plus for some customers, and all the new Power View functionality makes it quite an attractive web-based reporting tool for SSAS users. The per user cost of a Power BI licence might make it too expensive to buy if all you want are web-based dashboards and mobile support but customers who are already checking out Power BI for self-service will be more likely to buy because of this.


New features in Q&A were not mentioned in the keynote on Thursday, but on Friday afternoon I saw an interesting session that detailed some of the new functionality coming in Q&A later this year including the support for phrasing that is mentioned in this blog post. It’s clear that Q&A is getting a lot of love at MS and technically it is very impressive. I’m still not totally convinced that this is something people actually want or will use but I’m less cynical than I was. I also smell a lot of consultancy money in building and tuning models for Q&A if it does get popular (the session showed that there are a lot more features coming that will help with tuning). If I understand correctly, the output from Q&A is basically a Power View report which can then be edited manually if you wish; this means that Q&A should not be thought of as a standalone tool but as one of the ways that you can start to build a Power View report, and Q&A will benefit from all of the new features that are going to be added to Power View.


Apart from the time series forecasting, which is available now, ‘this summer’ was the most common response to all questions about when the new stuff would be available. Maybe all of these features will be released when the introductory pricing period for Power BI ends? Or in time for the Worldwide Partner Conference in mid July, as Jen Underwood suggests? Hopefully it won’t be in the middle of my summer holidays.

18 responses

  1. As someone who has had more than a passing interest in data mining and its adoption I thought I’d throw in my 2 cents – to me it strikes very close to the entire justification for the POWER move.
    Perhaps the low adoption has always been the delivery mechanism? After all, it was essentially left up to the IT end of the world to implement something that the business should have had their hands on in the first place. To be frank, that relationship never worked out because of IT’s instance on ‘Corporate BI’. Now a move to Excel (or POWER) solves this issue (as long as the message gets out).

  2. You mentioned that ‘business users don’t trust predictions when they don’t understand how they are made’. this is absolutely true. I had a case trying to move the SAS enterprise miner to Data mining to substantially save the cost. We set up the model and business case, we end up with very very similar results from both application with SSAS even spent less time. but they didn’t trust it because they don’t know how SSAS datamining actually made the results. End of story.

  3. The SSRS piece is the biggest announcement for me as well. I see this as being a great marriage between “stuff IT cares about” and “stuff the business cares about.” It makes Power BI a common platform that both sides would want to lobby for.

    Then IT doesn’t feel that they’ve been completely overthrown like I somewhat encourage here:

  4. Hey Chris,

    I agree that the biggest point is Integration with old-school SQL Server BI. The conflicting messages around self-service and corporate is very off putting for a lot of my customers and the inability to easily use Power BI with on-premise SSAS solutions is the biggest reason for them not to go Power BI. Did they share with you a roadmap or dates?!



  5. Love all the new features, unfortunately the monthly per user licensing model is still completely broken for us, and I assume for others (especially without E3/E4). I mean, $480 per user per year? At 200 users that’s $96k per year. In perpetuity. Power BI would be our most expensive annual software spend across all of IT. Paying per user per month is a very low resolution version of the cloud mantra “pay only for what you use”. Am I off my rocker?

    • I think you raise a very important point here, Chris. The licensing is a bit confused: ideally only a few users, the ones who need use Q&A for example, would need a Power BI licence and the rest should be able to make do with a regular O365 subscription, and I think that is Microsoft’s intent. However you are right there are plenty of reasons why this won’t work and why the only safe option is to buy a Power BI licence for everyone, which is horribly expensive. This needs a blog post…

  6. Pingback: Power BI – Musings in May 2014 | Big Bear BI Blog

  7. Pingback: BPOTW 2014-05-30 | SQL Notes From The Underground

  8. Pingback: New Power BI Features Shown At WPC | Chris Webb's BI Blog

  9. Pingback: New Power BI Features Shown At WPC - SQL Server - SQL Server - Toad World

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: