Articles for the Month of September 2015

Creating a Successful BI Project starts with Data Modeling

Recently I was helping someone debug an Analysis Services Multidimensional project, and didn’t come up with much to help the performance. Why? The underlying data model was completely unwieldy and the fix, which no one wanted to do, was to redo it completely. Having worked recently with a number of business analysts to migrate there Excel spreadsheets to Power BI to support the growing trend to Self-Service Business Analysis, has made me think a lot about what makes a project a success.  Self-Service BI has been hyped as the way that analysis can better do their job and not involve technical resources.  While I support the move to the Analysts being more involved in with the data to make good decisions using the data, these kind of projects still need experienced data professions help them make a the project a success. There isn’t a tool which can fix a project with a bad data model. The problems the analyst have are not so much with learning the tool, as Power BI was designed to be easy to use. The problem is with data modeling.

Reporting Views; Modeling for the Moment

A lot of business reporting is developed by using the following process, which you may find where you work. The database team can’t keep up with the report requests, so they create a number of views and provide business analysis with some tool, be it Report Builder, Excel or Access to gather the data to do reports. This method provides the ability for analysts who don’t know much about data modeling to create reports based on the information is provided. This process works for a while. As long as the data people need to do their jobs is provided, reports are created and the Database team doesn’t have to be involved. This whole methodology starts blowing up over time. Why? The reporting time starts to increase.

The Reporting Time Explosion

Once I was working at a company where the person in charge of doing the performance reporting went on an extended medical leave, and trained someone else on what was required to get the data and create the reports. She gathered data from this system and that system, added in some information on a spread sheet, ran some macros did some queries, updated some Excel spreadsheets and after that the reports were generated. This process required three hours every day to do this complicated series of task and a full week for monthly reporting. It took all of about two days for her replacement to be overwhelmed, and the task of doing the reports came to me. After a week, I had gathered all of the data together for the daily reporting and automated it, which took the daily reporting process from three hours a day to seconds of computer time. It took a couple of stored procedures, some SSRS reports and a new process for storing the data not in an Excel Spreadsheet, but in the application where it was supposed to be entered. By the end of the second week, the monthly reporting was completed as well. A task which took the majority the time person spent her day, was automated to button clicks in less than two weeks. Why? The task of gathering the data was given to someone who understood databases and data modeling. That’s the knowledge that is needed to set up a successful BI Project.

Business Knowledge needs to be combined with Technical Knowledge

To be an expert at something takes time and focus. There are only so many hours in a day, and if you are focused on spending those hours on creating technical solutions, you are bound to get really good at applying technical knowledge gained to solving problems. Likewise, if you spend all of your day looking at the data trying to solve business problems and answer questions about how the decisions made impact the data, you are going to get really good at analyzing business data. Tools help provide the ability to answer questions, which can be answered because the data model supports the type of analysis needed. To figure that out, someone who knows about data modeling needs to be involved to ensure the Self Service business intelligence project has a good foundational data model.  If that’s not there, it doesn’t matter what the tool is, the project won’t be successful.


Yours Always

Ginger Grant

Data aficionado et SQL Raconteur

How 2016 Releases of SharePoint and Excel Impact Power BI

There has been a lot of news recently from Microsoft on the upcoming changes to Power BI, Excel and SharePoint. Some of the changes may impact your decisions regarding report distribution.

SharePoint 2016 has no Power Pivot Gallery

Within SharePoint 2016,  Excel Services have been removed from the SharePoint Server.  The Excel Services didn’t go away, they were moved to the Office Online Server Preview, which means if you want to continue rendering Pivot Tables and Power View via the PowerPivot Gallery, you won’t be doing this from SharePoint. If you are wondering how you are going to be able to view Pivot Tables and Power View reports in Office Online Server Preview, you have a lot of company. Office Online Server Preview is not out yet. What Microsoft is saying right now with the Preview Release of SharePoint 2016 is Excel Services are not there, and if you want to use Excel Services, you will be waiting until there is a Preview of Office Online Server. If you are thinking of upgrading next year to the latest version of SharePoint, you will need to wait to see how to do that and still support Excel report distribution.

Decreased Excel Power

Excel2016 cchangesMicrosoft is eliminating the rampant use of the word “Power” when it comes to Excel tools.  I want to stress that none of the tools have been removed, just renamed. Power Query has been renamed “Get and Transform”. Power Map is now called 3D Map. Only two power words remain, Power View and Power Pivot and the use of the word Power has diminished with both of them. Power View is still there and called Power View but you have to add it to the ribbon to see it. The Data tab now contains the familiar Manage Data Model icon to open up the Power Pivot Window, which you will have to enable the add-in to make it work. The add-in is found in the Data Analysis Add-in group with no reference to Power. A number of new features have been added, especially when it comes to visualizations in Power View, which now has Sunburst, Waterfall, Histogram, Pareto, Box & Whisker and TreeMaps.

Forecasting is Back

Forecasting in Excel 2016I really liked the forecasting Feature in Power BI Office 365, which I wrote about when was disabled in December of 2014. Now it is back in Excel 2016. It works like it did before by looking at historical information in the past and using those trends to predict future values using variables to help you adjust the value for different conditions. This feature provides a simple way to view possible outcomes, which I think is going to be a very popular feature.

Power BI Decision Process

Next year’s releases are going to provide a lot of changes which may impact what your organization may elect to do in the future. A lot of these changes were rather predictable, such as Excel 2016 containing the new visualizations found in Power BI, but others such as the changes to SharePoint, may be a surprise. Since I do not work for Microsoft, I have to guess like everyone else what the changes will be, but I don’t think it is a big stretch to guess that Office Online Server may be cloud based. SharePoint 2016 product announcement describes the product as a Cloud-hybrid, and I am guessing the hybrid part may see your Excel documents on Office Online Server in the cloud. If this is a big deal to your organization, you may want to read the information about this really closely. I know I will and will be writing about them here.  Please feel free to subscribe to my blog to get the latest updates.

***For updates on this topic please see my more recent post Update on SharePoint 2016 and Excel Services

Yours Always

Ginger Grant

Data aficionado et SQL Raconteur