Governed Corporate Power

Best wishes and many Data Insights. 

Oleon, my client, is at the forefront of the growing market of natural Chemistry. As lead of the Analytics team at Oleon it is my desire to provide useful easy-to-read business driven reports.

In a previous article I’ve been pondering on the question SAC vs Power BI ao visualization.

Now, 1,5 year later we are still confident in the combination of Power BI on a 7.5 BW Analytics System and a 2.0 Hana.  And during the holiday season, I have the luxury of being able to look back.

We are using Power BI in the cloud, easily accessible, automatic login with corporate laptops. End-users and Self Service users only need a browser. For handheld devices there is also an app should the user prefer it over a browser.

Learning Curve of Power BI

As with any new tool, there is a learning curve.  For end-users to self-service users this is close to instant. Almost all the questions I get from this side are functional, as it should be. Adding an information page or a link to definitions is a good help here.   

For Power users and developers the plus of Power BI is that ‘the first learning bump’ is easy to take and fills your head with grand dreams and a hunger for more. The second learning bump (programming with DAX and modelling for advanced visuals/calculations is much higher. 

Modelling

Important to know/learn is where to do which coding/modelling/manipulation and to get everybody in the team on the same page.  This takes time and there is no easy answer.

With BW on HANA you can model/code on the source system (usually SAP), in BW layers (the old LSA+, which with HANA you try to minimize to the essential and virtualize the rest), in HANA Calculated Views and finally in Power BI itself. (personal note: I prefer to link Power BI with HANA Views, I find it more stable/mature).

Classic SAP consultants will model too little in Power BI and classic Microsoft consultants vise versa, both making it far more complex then needed.  Both need to understand each other’s strengths for optimal results in the most efficient way.    Myself with a strong SAP background, I had to learn that modelling with master data, next to transaction data in Power BI has benefits in certain scenario’s (filter context, sumx, ..)

This approach requires to shift more logic from the reports (BEX Queries) to the HANA Views.  Only Aggregation depended on logic and logic for visualization is still needed in your reporting layer. Plus side is that this approach makes it easier to switch to a new reporting tool(s) in the (far?) future.

Source systems

Power BI will let you happily combine data coming from all sorts of source systems on premise or not also production and development, so you will need clear guidelines to prevent blunders (reporting dev data on a production system). We of course use ‘import’ method, since we do not use an SQL layer in Azure or other.

A follow up on this is flat files.. Classic SAP approach is to upload your files in an aDSO, transport this set-up to production, and publish this to reporting.. (2 to 4hours to set up) Advantage is that this process is more IT-governed with more fail-safe in case of data entry errors. The alternative is to place flat files onto OneDrive (easier to set up) or a network drive and refresh the data. This set-up is way faster, and typically takes about 5minutes. Refresh of data can also be automated via the Gateway. This second approach can be applied for files that are and will be only relevant for this one report and are maintained by IT.

Visuals

In your first hours of using Power BI you will quickly dive into the marketplace for EXTRA fancy visuals.. only to find out later that the better visuals in there cost money. Much later you discover that these marketplace visuals have caveats.. they lack e.g. language translations, behave different for tooltips, don’t work the first days after a new version .. or other.  In a corporate environment, you will mainly want to stick to the visuals provided by Microsoft.

Governance

Governance is absolutely essential, the playful joy of Power BI almost makes you forget this potentially catapulting your company back to 1980. Governance is a chapter on its own. My wonderful colleague Sander from Lytix wrote a short take on this.

His blog can be read on this link: https://new.lytix.be/uncategorized/governed-self-service-a-power-bi-rollout/

Final thought

Power BI has a low threshold and still has the option for complex constructs if needed with an excellent distribution platform in the cloud. There is a large user base and a lot of information, youtube videos and many more. Visualisation is the tip of the Analytics iceberg, but it is the part that users see and interact with. This layer has to be and is easy to read/pinpoint opportunities and trouble-free/time saving to use. Inviting users to look more at their data.

Blog by Carl Goossenaerts