To be able to discuss analytics in a structured and comprehensive way, we have developed a simple framework that describes the main components and responsibilities in an analytics solution. We have given it an equally simple name: the Five-Layer model.

Figure 1.1: The Five-Layer model

The first and lowest layer, Connect, is the starting point of analytics: if you want to analyze data, those data must come from somewhere. They could reside in Excel sheets, text files, large business databases, or somewhere on the Internet.

These raw data aren't typically in the right format to be able to analyze, especially when they come from different sources. Therefore, you need to Prepare your data. Preparation can take many forms, like imposing certain data types, transforming data, building data history, or matching data based on key attributes.

Creating nice and clean datasets in the Connect and Prepare layers can take...

Organizations are becoming ever more data-driven. An organization that assesses performance on key performance indicators, or KPIs, uses dashboards reflecting the status for each KPI. These dashboards are typically highly standardized: the organization has reflected on the business strategy, business processes, how to measure the KPIs, and how to report on them. An automated dashboard with KPIs is relatively stable and will not change much, and is typically built and maintained by a central IT department or BI competence center.

The next level of being a data-driven organization is that every decision taken can be based on insights from relevant data. This means that the need for a more dynamic form of analytics arises, one that can answer ad hoc questions. This form of analytics is marketed as self-service BI and promises that everyone can build a BI solution without the help of a central IT department. From the Five-Layer model, it is clear...

In an analytics solution based on the Microsoft platform, DAX is used in the Analyze layer. DAX lives inside analytical models as the formula language to define calculations and other logic. In fact, models and DAX are really two sides of the same coin: the design of the model impacts the complexity of the DAX code, and your skills in DAX determine your model designs (we will elaborate on the core concepts of data models in Chapter 1.2, Model Design).

The power of DAX is in its strong data aggregation capabilities. The DAX language contains many functions and constructs for defining a variety of aggregations to generate results that lead to the insights needed. In the past, many types of aggregations could not be created directly but had to be implemented through specifically preparing data. For instance, computing a year-to-date sales total can be done in DAX with a single function, while in Excel or traditional reporting tools, separate...

You have multiple options to choose from as your tool to work with DAX, depending on the target platform for the model:

• For a Power Pivot model, you use Excel with the Power Pivot add-on.
• For a Power BI dataset, you use Power BI Desktop.

  It is interesting to note that there are in fact three versions of Power BI Desktop. One can be downloaded from the Power BI website, powerbi.com. The second is installed from the Windows Store and is automatically updated just like any other app from the store. When you realize that Power BI Desktop gets a new release nearly every month, these automatic updates are a real plus, although it can sometimes be annoying when a new release changes things you do not expect. If you want to, you can install both versions on the same computer. The third version of Power BI Desktop, which can be downloaded from the Power BI website as well, is a special edition for use with Power BI Report Server.

  ...

In discussing the Five-Layer model, we already briefly touched upon the importance of visual reporting. But effective visual reporting is only possible through the power of DAX-based models.

Creating visual output is paramount to generating real insights versus a lot of information only. Just take the example below:

Figure 1.3: Some sales data in a table visual

Can you spot the problem or opportunity this company has? If so, you are very good at numbers! Most people, though, are much more visually inclined. The picture below provides the same insight in a visual way:

Figure 1.4: The same sales data presented more visually

It is obvious now that one of the SKUs is doing significantly better than the others. That is an interesting and valuable insight: if we can make the other SKUs reach the same performance, the overall results of the company will dramatically improve. This insight also leads to something else...

Power BI models and DAX enable a different way to develop BI solutions with much deeper involvement from businesspeople. This leads to solutions that deliver insights that optimally add value to the business.

The traditional, IT-centric way to create BI solutions is to start with connecting to data sources and preparing data. At first glance, this sounds like the right thing to do; after all, we need good data to have good and valuable insights.

Figure 1.6: The traditional BI solution development approach

This approach is typically materialized in the development of an enterprise data warehouse. The idea behind having a data warehouse is to store all of the organization's data in a single place, and to use that as the foundation for all reporting.

It should be clear that this can be a major undertaking, as organizations have many different systems, with lots of different data in them. The data warehouse is traditionally...

So far, we have focused on what is needed to go from "raw" data to insights, and the role of DAX-powered Power BI models in this process. We have seen that business value does not come from merely connecting and preparing data, but that it is the insights from visual, interactive reports that provide value.

However, no organization will get better from just staring at nice-looking reports. What really matters, of course, is what you do with the insights. In other words: you need to take action to reap the benefits of a BI solution. You will also want to measure the effects of these actions, either in an automated way or by letting users enter feedback into some kind of system. The result of that is, again, data.

This leads to a cycle of digital transformation, or data-driven business improvement:

Figure 1.8: The digital transformation cycle

While Power BI shines in the process of going from data to insights, it is not...

In this chapter, we discussed the field of business intelligence and the central role of analytical models in modern BI solutions. Power BI models are ideally suited for use as such models, not least because of the power of DAX.

You have learned about two capabilities of DAX that have a profound impact on the way BI solutions can be designed and developed:

• DAX enables sophisticated calculations for a variety of aggregations of data; aggregations that, in the past, needed a lot of preparation of customized data. Therefore, DAX allows for shifting the focus from data (with all the tedious work involved) to the logic to generate business insights.
• DAX as a language is set up in a way that allows businesspeople who are mostly familiar with Excel to work on the BI solution themselves, to varying degrees. This means that much more alignment with business priorities can be achieved.

As Power BI models and DAX are two sides of the same coin, it is important...