3 things I learnt from implementing Power Bi

I used to perform a lot of my analytical work in Excel, which was the wrong place for sure.?But not knowing any better, I was stuck, crunching.

Then, my wife showed me this talk by data scientist Hans Rosling, which combined great public speaking skills, a well-researched topic, and compelling visuals. That was a shock.

At that time, I had long heard that automation was coming for excel-based roles. And although Excel and PowerBi are certainly not inter-changeable, the talk inspired me to get into the latter.

3 years down the line, hundreds of hours later (if not a solid thousand), I wanted to put in a nutshell some key learnings from there, how I grew, and share the hidden truth that some Power BI owners might experience but not say.

Before we start, let’s make a clear point that excel-based roles should learn Power Bi, or at least PowerPivot (but really, you should skip to Power BI directly). Why? If?you want to deliver high value insights to make decisions, you cannot afford to crunch excel infinitely.

If you start developing PowerBi for you own use, you will save your time by automating data sourcing, and easily be able to find patterns, answer your questions, and present your findings.?Get excited and read-on!

?1.??????Programming languages

?I’ve never been a developer, just had some VBA experience but that was it. Power Bi will get you to code for the better. Here is a list of what you will encounter:

?DAX : MUST HAVE

Power Bi speaks DAX internally. This is the language you will use to develop your measures (your “ KPIs” ). DAX is simple to read, and honestly simple to learn. It has built-in formulas that will guide you to do what you want, and “intellisense” will support your programming. For a solid take on DAX, these courses are gold:

SQL: OPTIONAL

SQL will nicely fetch (slice and import) your data to PowerBi. Chances are extremely high that your organization stores data in relational databases, and SQL will be your friend to extract data. The syntax is close to English and fun to learn.

Spoiler: if you don’t know / don’t have time to learn SQL, PowerBi data model is equipped with a built-in functionality to create the relationships you need. In other words, just import your data table by table from your database and do the connecting and filtering work in PowerBi user interface.

?M: OPTIONAL

No James Bond pun here. The language is called M (Data Mashup, or Data Modelling). Once you have connected your data to Power BI, you may want to manipulate it before using it with the rest of your data model. Power Query (built-in add-in in Power Bi and Excel) will help you do this, using “M”.

?The good news: the user Interface is so good that you will hardly ever need to code anything.

But in some specific cases, you may have to write a few lines of code.

This course helped solve 100% of these nasty situations. Highly recommended, but if your data is simple and clean enough, you may never need it.?

?Python: OPTIONAL

Long story short, Python visualization libraries extend beyond these found in PowerBi and can be useful to know in some cases. But again, 99% of visualizations you want to render have already been developed as visuals in Microsoft Store, ready for you to download for free.

Python can flex big muscles going into machine learning and render the results in PowerBi, but let’s keep it for a separate topic as it is not a must to create successful PowerBi visualizations.

??So, if you’re insanely good at Excel but shying away from code, fret not.?Have a play with some of the ready-made datasets for PowerBi (some of the courses indicated above also come with their datasets to follow and practice) and learn by doing - this is the only way.

?

?2.??????End user usages

?Map out the most urgent needs first

I have worked long hours to make the best possible product for myself, my stakeholders, and colleagues.

Funnily enough, the pages (reports) that I used the most took me 1 week to develop, when the ones I made for my colleagues took me 3 months. The truth is: they use it the same way I do: ?only sometimes, with a specific question in mind, not really minding the “colors” etc.

?As product owners, we are the most passionate in this project. However, stakeholders and recipients have other plates to spin and will not be as passionate as we are. They will only come to find information relevant to their needs, when they need it, which may not be everyday (even though you developed that most-wanted “daily report”).

?

Mobile version

Given that we spend on average 5 hours a day on our mobile devices, I also developed the mobile version of most of my pages, which is basically duplicating a lot of work.

Most of my stakeholders open PowerBi reports between 3-7x per month, and only on the web portal to find what they need.?Takeaway: make the mobile App a part of a Phase 2, not a Phase 1. Chances are you might never need to develop the mobile version.?

??

?Don’t go all out:

I shipped a 95% final product, when a minimum viable product (MVP) would have solved the same problem. Lots of time and effort spent learning, but not really adding value. Beginner’s mistake!

I learnt after this the ropes of AGILE Project Management, which outlines this clearly.

?A former colleague reminded me of this reality: there are business problems to be solved, and the tools most of the time don’t matter. Build Power Bi with this focus in mind: SOLVE.THE.PROBLEM.

The rest (colors, display…) can wait.

?Takeaway:

A good practice is to develop 1 Power Bi page, containing 2-3 representations max. Showcase it, let it sink for a week, ask for feedback, and then build. Don’t go too crazy with the layout. Less is more.

Absolutely worth having a 10min read at basic AGILE principles?or take a short course on it. Highly recommend this one, lots of “I should have!” moments

?

3.??????Maintenance and upgrades

Power Bi requires some maintenance and will break from time to time. The worst that happened thus far was when I had to rebuilt half of the app from a previously saved version because my PBIX file was declared – from the day to the next – “corrupted”.

Documenting the data model, measures, pages is extremely time consuming. And there’s no escape, you must do it while you develop your features, otherwise it’s forgotten. I use OneNote for this. But depending on the complexity of your project, you may want to consider other tools.

Shipping new developed features is easy, but there is always this fear of working on a “production .pbix” file. To my knowledge PowerBi does not provide separate development>test>production environments for developers, so be mindful of always saving your work (the versioning / restoring tool in windows 10 saved me a few times).

?Finally, make sure you get a seat at the “IT roadmap” table to understand how your data sources might be affected (example, if your organization shifts from AWS to Microsoft Azure for data storage– which does not happen overnight anyways - make sure you know ahead of time to update your model sources).

?

So, is the destination worth the journey?

One word: Yes. A million times.

?Learning PowerBi will not only expand your mind and give you “ah-hah” moments every day.

It will help you stay relevant in your role, by adding value, on a real-time basis.

If you get stuck, the support community is large enough to find answer to any problem you will encounter. However, for a structured approach, do check out these guys, absolute reference in this field:

?If you already have some experience using Power BI, please do share how you got started, what were the first quick wins you harvested, how knowing it changed your role and value in your company.

?Good luck with your project and do comment or share if this helped you!