Last Sunday was the annual BI Bake-Off, arguably the BI industry’s most highly anticipated, public competitive BI/analytics vendor demo showdown of the year. The mega-popular event founded by Cindi Howson is now held at the Gartner BI Summit. In a bittersweet battle of top BI vendors that included brilliant minds Francois Ajenstat, Vice President of Product Management from Tableau, Josh Good, Director of Marketing from Qlik and Chris McNabney, Senior Vice President of Product from MicroStrategy, I was honored to showcase the latest Microsoft Power BI offerings. I do believe Gartner will be posting a blog overview soon. In the meantime, here is a summary of my experience and demos.

Getting the Invite

Six weeks ago, I received an exciting invitation for Microsoft to participate in the BI Bake-Off. I was thrilled to join in on the competitive fun. After accepting the invitation and working through internal chatter on the matter, I received instructions on the required scripted demos and how to find the College Scorecard data sets to use for the build. Upon an initial scan of those requirements that intentionally included missing features across all vendors, I knew that I had to pull in a custom data viz resource and send out feature release escalations to ensure a truly fabulous showing. Apparently I was not alone in that scramble.

Several vendors on the stage released features specifically for the BI Bake-Off event! You’ve got to love our deeply competitive spirit.

Planning the Demos

Marketing 101… From there, I studied the scenarios, anticipated what competitors will highlight and then chose what benefits/features will effectively differentiate my offering to position Power BI in the best light possible regardless of the scripts. If you don’t have strong functionality in an area, razzle, dazzle and distract with other stuff. Do you think I am kidding? Well, I am not. I did call out the guys for playing that card in the data prep section this year. They did the same thing last year in other areas. If vendors don’t show the required demos or are missing features, they do get called out on it by other vendors on the stage or by the moderator. That is all part of this totally enjoyable competitive exercise that allows customers to see BI solutions in a fairly comparable, side-by-side manner. If you have not been to a BI Bake-Off before, I highly recommend going to one if you can.

Walking through Microsoft’s Demos

Each vendor was given the opportunity to introduce themselves and present a summary of what makes them unique. For Power BI, I shared the tagline “Experience your data. Any data, any way, anywhere”, painted the future vision of innovation and then went on to quickly iterate through the following:

  • Weekly pace of feature releases
  • Natural language Q&A and voice Q&A with Cortana
  • Proactive alerting with user defined thresholds on mobile apps
  • Real-time analytics and live streaming dashboards
  • Rich data prep with 200+ point-and-click transformations plus M function library
  • R script integration and R data visualizations
  • Open source custom data visualizations
  • Unified dashboards containing Power BI, Excel and Reporting Services
  • Office integration with Excel, Delve and Collaboration features
  • Interactive data storytelling with Sway
  • New Power Apps for end-to-end insights with workflows and write-back
  • Truly global community/ecosystem of Partners and Developers

All demos were supposed to help answer the key business question: Which states have the best schools for academics, value, and earnings potential?

In first required demo, we were instructed to show two geographic maps by state (highest SAT averages, highest earnings), include a trend over time for average SAT scores or for declining enrollment numbers, rank top SAT scores, or biggest percentage decline in enrollment and provide a cross tab comparing several schools for scores, cost, earnings, debt. The script required filtering with elements and sliders, resorting data, and drilling down to detail from state to individual school. We were allowed to get creative. Thus I chose to go with a March Madness theme.

Demo 1

In this demo, I leveraged a custom data visualization for the slider and cool college logo slicer. I also sprinkled in a dynamic infographic trick that I recently learned from talented Miguel Myers of the Power BI Community. You’ll also notice the trend chart with banding. For that one, I used ggplot2 linear regression. For rankings, I used DAX tips from the Rob Collie.

In the second required demo, I showcased the cooked in Power Query features within Power BI Desktop. Secretly I was hoping that Miguel Llopis on our team could wow the crowd with this particular demo but this year we were only allowed to have one presenter for all demos. Thus I did not give data prep the justice it deserved – especially since I still am having trouble seeing tiny menu text on my laptop.

Data Prep

For this one we were instructed to start with a basic flat file and show how someone creates an initial visualization with filters. I showed our library of data sources, chose specific columns to reduce size of data loads, then designed a hierarchy and added a data visualization with a drill-down. Then for the next part of data prep, intelligent data type detection, automatic join detection, data cleansing, transformation and blending with an external file, I ran out of time. Uuuugh!  This is also the area where several other vendors decided to go completely off script. I have no idea what they did but recall being totally surprised by it.

For demo three, dashboard design and story telling, the requirements included showing how you created the dashboard in the first demo with two visualizations and one filter that works across the visualizations. This demo also required showing how snapshots and external content can be woven in into an analysis of findings. Admittedly, I needed to get creative here. I chose to show Power BI dashboards, widgets for images, external iframe content, an incredible custom data viz Leaflet Map made by James Dales of Altius, sharing dashboards and also the fantastic Sway integration with Power BI for data storytelling.

March Madness


For the next required demo, we had to show subscriptions functionality. We needed to demonstrate dashboard scheduling on a daily basis or event basis with output options: native, Excel, PDF. For a few vendors, subscriptions is not a high priority feature. With modern mobile BI apps and offline access to reports, there is less need than there has been in the past for it. Regardless, it was a required demo that I needed to figure out. I initially tested scripting Power Shell commands, automating keystroke commands (yes, I know… a wild stretch) and then I remembered a partner utility from Power Planner called Power Update. Power Update ended up being totally awesome for the task of sending refreshed Power BI Desktop files in PDF or native format. Power Update secured us a great audience score! If you are not familiar with it, be sure to check it out.

Power Update

The next required demo centered around governance features and role based data security. For this one, I needed to show how a self-service BI data model can be shared and promoted to a governed data model. I also had to demo how security can be set by row level by person and state. In my walk-through, I first shared how Power BI models can be upgraded to governed Analysis Services models and then have dynamic data security applied all the way down to cell level if desired. Then I shared a surprise second option where business users can apply role based security that does not require SQL Server or Analysis Services – more info coming soon on that one.

Row Level Security

In the very last demo, each vendor could show innovations or differentiators that set our solutions apart from one another. This is where I went rapidly through the long list of what I feel makes Power BI powerful. If you could keep up with me and my beloved Diet Mountain Dew, I literally whipped through the following demos:

  • Quick Insights with a few odd findings not caught in manual data analysis
  • Natural language Q&A and mentioned voice Q&A with Cortana
  • Unified dashboards containing pinned Power BI, Excel and Reporting Service
  • Office integration with Excel, Delve and Collaboration features
  • Interactive Excel pivot charts and tables in Power BI
  • Content Packs with pre-built solutions
  • R data visualizations and custom data visualizations
  • Real-time IoT streaming dashboards of the Seattle 520 bridge

Massive Sigh of Relief

When all was said and done, I was incredibly relieved. Although I did not achieve top audience ratings in all of the area demos. No one else did either. My tiny three person cheering section that included my two super supportive analyst relations team members confirmed those findings. I had held my own on that stage in front of an executive decision maker and BI pro crowd of ~500 right beside highly intelligent, savvy Vice Presidents of Product Management and Chief Marketing Officers. It is an experience that I will never forget and I will forever cherish.

BI Bake-Off

Someone tweeted this year that all of the tools looked the same. Back in October 2013, I mentioned an inevitable analytics market commoditization would occur in the data visualization space. We are indeed starting to see that prediction become a reality along with my more recently predicted BI wars that is putting increased pricing pressure on the premium vendors. I personally know where there are significant depth and breadth differences across these offerings but… I am an expert in this area. From a non-expert perspective, many of these tools sure do look and feel similar to one another today.

Looking Ahead

Citizen Data Scientist

It is going to be far more interesting next year to see how the vendors choose to differentiate and expand core offerings. While at the Gartner BI conference, I saw more automation of predictive analytics and text analytic content descriptions being baked into various data discovery solutions like BeyondCore. Yseop showed me a demo of analytic text automation within Power BI Desktop. Qlik now has Narrative Science added in. Tableau has forecasting features today. SAS, SAP Lumira and TIBCO Spotfire all have had predictive features for a while now within their data visualization tools.

I have not yet seen enough prescriptive analytics cooked into data visualization tools for what-if, optimization and simulations to help the business users, or shall I say citizen data scientists (buzz word to add to your resume), get the most value from applying predictive analytics. Frontline Systems and Alteryx are still my go-to solutions for prescriptive today. Only a few vendors seem to be at that more mature phase of analytics including one of Gartner’s Cool Vendors called APT Applied Predictive Technologies.

Oh no, it is already 3:47AM. How did that happen again? I promise to share more news from the Gartner BI event, across our ever changing analytics industry and on a personal note soon. For now, I must sign off and get a few hours of sleep.