Would you leave your finances in the hands of someone who knows accounting software (Contaplus, sage or any other), but who does not have the minimum financial and accounting fundamentals? Surely not.
Why then do you use Power BI without the minimum basics of dimensional modeling?
Power BI is a data model-based tool, but it’s very common to work with it as if it were just another calculation and reporting tool. It is essential to change the chip. I will develop and discuss it in detail in this article. And, of course, I will propose a solution????
Mistake no. 1 mistake with Power BI is “not having a data model that adequately answers business questions.”
I have been designing analytical solutions with this type of tools for 20 years and this common denominator has always been maintained, the importance of the data model and the identification of many problems in it, which my clients a priori linked more to calculations and visualization.
Problem review: calculations and visualization
On many occasions, due to lack of knowledge, we assume that a database model designed and optimized to perform many concurrent writes and reads is also optimal for queries and analysis, but this is not true. We need a query and analytics optimized model to answer our business questions.
Let’s see a practical example that I encounter much more frequently than I would like and which was the fundamental reason why I write this article.
I’m reviewing an already developed BI solution and they tell me they are having problems with a calculation and/or a visualization. From there I start to investigate, and I often come to the conclusion that the problem is not there, but in the Data Model, in the foundations on which the solution is based.
The root of the problem: the data model
We begin to review the model and, above all, I find myself with a load of data that maintains the same structure present in the source. If for that model I need to get data from 17 tables and 5 files (csv, excel, etc.), I find a data model in Power BI that contains a total of 22 tables. That is, for each source table, I create a table in the model. And for each file I also create a table in the model. Total of the 22 tables mentioned.
Sometimes it’s not that extreme, small changes, calculations, conditions and groupings are made.
In both cases the underlying problem remains the same: the #1 mistake made with Power BI, and in general, in many BI solutions: “not having a data model that adequately answers your business questions”.
Whether you use Tableau, Qlik, Power BI or any other solution, this problem is common for all.
Example of improvement of a solution created by one of my students. Error no. 1 salesman with Power Bi
I literally put the contents of an email that David sent me, after having followed a premium course with me:
Subject: Evolution of the dashboard thanks to the Power BI course
«Good morning Salvador,
I’m writing you this email because I was so struck by the situation that I think I should share it with you out of curiosity. The other day at the end of the course, in the round of questions I told you that in my specific case I already had a dashboard created in Power BI (when I didn’t have the mastery of the system that I have now thanks to the course). I’m currently starting to put together the new dashboard for the company and will show you the before and after so you can evaluate for yourself. It’s true that I loaded all the tables because I didn’t know the relatedness of the tables and the relationships between them, that’s why the chaos was greater but still it’s curious????
Thank you sincerely for everything.”
This is how David went from a model with 20+ tables to a model with 4 tables, containing all the information he needed and answering his business questions much better.
It is not my intention to propose such a comprehensive change or training on Power BI. I simply want you to become aware of the problem and the need you have to create data models that adequately answer your business questions.
May you take a decisive first step.
Improved Power BI data models
To carry out this transformation of the source data, designed so that transactions are executed as quickly as possible (many people have to simultaneously read and write to the same database), you need to learn dimensional modeling techniques, and this is where I want to help you.
For this reason I created a course designed specifically for learning the foundations of Dimensional Models. With it you get that theoretical basis; so important and so scarce, usually due to lack of awareness, because no one has shown you that there are a set of defined and documented concepts and techniques that you need to apply.
You’ve heard of… star schemas, snowflake schemas, facts, dimensions, data history issues, data marts, data warehouses… and you’re not clear what each of these concepts is?
Well, I’ll tell you in other words:
Have you heard of… debits, credits, entries, notes, balances, assets, liabilities… and you don’t know what it is and yet you manage the accounting of a company? Certainly not, those who manage the accounting have these concepts very clear.
It’s time to gain this knowledge and dramatically improve your analytical solutions. I will help you.
Access the Course
If you want to know about errors 2 and 3 I invite you to download this guide:
Power BI: the 3 most harmful mistakes and how to avoid them
I hope you find it useful and improve your results with Power BI.
Leave a Reply