Welcome to your second lesson. In our previous session, we explored the four main types of data analytics, giving you a high-level map of what your team can accomplish with data—from describing the past to prescribing future actions.
Now, we'll zoom in on the fundamental vocabulary used to build those analyses. This lesson will define five critical terms that are the building blocks of any analytical discussion or dashboard: KPIs, metrics, dimensions, measures, and granularity. Mastering this vocabulary is essential for you to understand your team's reports, ask intelligent questions, and clearly articulate business requirements. Think of these as the essential nouns and adjectives of the data world.
Metrics vs. KPIs: Tracking vs. Targeting
Let's start with two of the most frequently used (and often confused) terms: metrics and Key Performance Indicators (KPIs). While related, they serve different purposes.
A metric is simply a quantifiable measurement. If you can count it or measure it, it's a metric.
- Number of website visitors
- Sales revenue per day
- Number of support tickets created
A Key Performance Indicator (KPI), on the other hand, is a specific, crucial metric that you have chosen to measure your progress toward a strategic business goal. All KPIs are metrics, but only a select few metrics are important enough to be designated as KPIs.
The following video provides a very clear explanation of this distinction and offers a simple framework for choosing good KPIs.
What are KPIs and Metrics? | Data Fundamental for Beginners
This video from "Alex The Analyst" breaks down the difference between metrics and KPIs with simple, practical examples.
Watch the entire video. Pay close attention to: The initial definitions of a metric and a KPI. The framework for choosing a good KPI, which involves asking about your goal, the best metric to track it, and whether that metric is actionable (the framework).
In short:
- Metrics measure processes and activities. You might track hundreds of them.
- KPIs measure performance against key objectives. You should focus on a vital few.
For your team, they might track a metric like "average query execution time." But the KPI for the business might be "ensure 95% of customer-facing reports load in under 3 seconds." The metric informs the KPI.
Dimensions and Measures: The "How" and the "What"
Now let's look at how data is actually structured for analysis. Nearly everything you see in a chart or a report is a combination of measures and dimensions.
-
Measures are the numbers—the quantitative values you want to analyze. They are the things you can do math on, like
SUM,AVERAGE,COUNT. Examples includeSales Amount,Quantity Sold, orPage Views. In a database table, these are often the numeric columns. -
Dimensions are the categories—the qualitative attributes that describe the measures. They provide context and are the "slices" you use to break down your measures. Examples include
Product Category,Region,Customer Name, orDate.
Given your background with databases, this concept should feel familiar. In SQL, when you write a query like SELECT region, SUM(sales) FROM orders GROUP BY region;, you are aggregating a measure (sales) by a dimension (region).
The next video explains this fundamental concept using a simple dataset.
What Are Metrics & Dimensions? | BI For Beginners
This video from "Learn BI" provides a clear, practical demonstration of what metrics (measures) and dimensions are and how they work together to create visualizations.
Watch the following segments: The definition of metrics, or quantitative data points. The definition of dimensions, or qualitative data points. An example in Excel showing how to identify metrics and dimensions in a table. A demonstration of why dimensions are essential for visualizing data.
These concepts of measures and dimensions are central to the design of analytical databases, often called data warehouses. In that context, measures are often referred to as facts, and they are stored in fact tables. The descriptive attributes are stored in dimension tables. We will explore this "star schema" design in a later module, but for now, it's useful to know where these terms originate.
To get a slightly deeper, more technical perspective on this, the following article explains the concepts from a data warehouse design standpoint.
What Are Facts and Dimensions in a Data Warehouse?
This article from Redgate gives a concise, database-centric explanation of fact and dimension tables, which store measures and dimensions, respectively.
Please read the following three short sections: Under the heading "Dimensions Versus Facts," read the definitions that distinguish between the measurable events (facts/measures) and the reference information (dimensions). Next, read the section Defining Dimension Tables to understand their characteristics. Finally, read the section Defining Fact Tables to see how they contain the quantitative measures.
Granularity: The Level of Detail
The final term for this lesson is granularity. Granularity refers to the level of detail represented in a single row of your data.
- High Granularity (Fine-grained): Each row represents a very detailed piece of information, like an individual product in a single customer's shopping cart (
order_line_item_id). - Low Granularity (Coarse-grained): Each row represents a summary of information, like the total sales for an entire day (
sales_per_day).
The process of moving from high to low granularity is called aggregation. For example, summing up all individual transactions to get a daily sales total is an act of aggregation.
Understanding granularity is critical because it determines the types of questions you can answer. If your data is only available at a daily level of granularity, you cannot answer questions about which hour of the day is busiest. Your team often has to make trade-offs between storing highly granular data (which is flexible but requires more storage and processing) and pre-aggregated data (which is faster for specific reports but less flexible).
The Redgate article you just looked at also has a brief but important section on this topic.
What Are Facts and Dimensions in a Data Warehouse?
This is the same article as before. We will now focus on the concept of granularity.
Please find the heading Granularity of Facts and read that section. Note the key point that the grain of a fact table is a critical design decision.
Conclusion
In this lesson, we've defined the essential building blocks of data analysis. These terms are not just jargon; they represent a precise way of thinking about and structuring data to answer business questions.
Here are the key takeaways:
- Metrics are any quantifiable measurement, while KPIs are the critical few metrics tied directly to business goals.
- Measures are the numeric values you perform calculations on (the "what," e.g., sales).
- Dimensions are the categorical attributes you use to slice, filter, and group your measures (the "who, where, when," e.g., region).
- Granularity defines the level of detail in your data, determining the depth of analysis possible.
As a manager, you can now use this vocabulary to have more productive conversations with your team. Instead of asking for "the sales data," you can ask for "the sum of sales (measure) by product category and region (dimensions) for the last quarter, at a daily level of granularity (granularity)." This clarity is a superpower.
In our next lesson, we will step back and look at the bigger picture again. We'll outline the typical end-to-end workflow of a data analyst, showing how they take a business question and use these concepts to deliver an answer.