No wobbles: Teaching data modeling

When I was first starting to teach Power BI, one of the hardest concepts to teach was data modeling. A good data model (and by “good” I mean well thought out) is a work of art, and as such, takes planning and thought. Unlike using a Vlookup formula in Excel (for example), establishing a relationship between two tables in a model has implications for other relationships in the model, either existing or proposed. To add insult to injury, your data model directly affects your visuals. I tell students that if you have a problem with a visual, 95% of the time the issue causing the problem is in the data model. This means that even in the beginner Power BI classes, a good trainer introduces the concept of data modeling.

Unfortunately, data modeling is a concept that is unfamiliar to most business users. It was new idea to me when I first started ‘playing’ with Power BI, and it was a significant barrier to entry for me. I had no prior database training and I was relying on my Excel knowledge to provide the scaffolding. The good news is that there are some methods and concepts in Excel that can help accelerate a Power BI newbie’s learning curve, but at the end of the day, data modeling skills are essential. That’s why I make sure to do two things in my training:
1) Explain
2) Reassure

Let’s start with the second one first because it is the most important. Most business users are several years past the end of their formal schooling. Gone are the days when they were regularly presented with an intellectual challenge that they have to master. It’s easy for adult learners to get rusty on all the skills they used to master complex ideas. I like to remind my class participants that they retain this ability. But I like to go further than this rather anodyne reminder. I contend that most people naturally make the connections necessary to fathom how a data model works. I use the example of two lists on two different pieces of paper. If you want to find out which items are on the first and second list, you compare the two lists side by side, using one of the description items as a comparison. So if you are comparing a list of all the appliances you have to a list of the best appliances (by price or rating), you could go down your list by brand, and then look at the best appliances by brand. In this case, you are making a relationship between the two lists based on the brand description. Or you could look at the type of appliance you have (dishwasher, refrigerator) and then compare that to the list of best appliances by type of appliance. Here, you are making a relationship between the two lists based on the type of appliance. This isn’t intuitive behavior (at some point in our lives we learn to do this), but it is something almost every adult does on a regular basis. So I am not teaching them a new skill. I am showing my students how to apply a skill they already know in a new scenario.

Now let me break down the “explain” part of the equation. A key role of a Power BI trainer is to introduce the vocabulary of the discipline to the newbies. Part of the barrier business users experience is the use of new terms to describe tasks or Power BI’s behavior. The hardest one for me was “ETL”. It had two things going against it: 1) it is an acronym and 2) it is new terminology that didn’t seem to apply to what I was doing. (Just for the record, “ETL” stands for Extract, Transform, Load.) I start with this acronym, deconstruct it, and go on to other terms that are opaque: relationships, cardinality, primary and secondary keys are just a few that come to mind right now. Once you know what these terms mean, they unlock a lot of other training materials, but without them, your learners will be like Alice, wandering around in Wonderland, bemused and confused.

Remember when you learned to ride a bike? (Apologies to readers who have never had this pleasure and privilege.) Until you did it for the first time, no one could have convinced you that the forward motion of the two wheels will be sufficient to balance you. You probably had someone holding the bike steady and encouraging you to trust and go….Hopefully not someone like me. I yelled at my daughter until she was so mad she forgot to be afraid! Parenting fail. Unlike my parenting effort, as a Power BI trainer, I strive to be encouraging and reassuring to my students. But at the end of the day, the student has to make the leap of faith that s/he can learn data modeling, at least well enough to use Power BI to get from here to there without wobbling.

Photo credit: Jordan McGee on Unsplash