Metrics. The Building Blocks of Modern Self-Serve BI

Metrics are key to modern, effective, self-serve Business Intelligence (BI) analytics. In this article, we’ll describe how to move from rudimentary data tracking in spreadsheets to a sophisticated, self-serve system using metrics. We’ll dig into the challenges of traditional BI models, highlight the benefits of a metrics-centric approach, and provide practical steps to help your business make better, more informed decisions with metrics.

From Excel chaos to self-serve intelligence

Every business, big or small, has to start somewhere. Often, this journey begins with a simple idea and a lot of enthusiasm.

Excel spreadsheets and enthusiasm

Imagine two high school friends inspired by an idea: sports equipment for pets! They dive in and start tracking data. Initially, it’s all in Excel. They manually cut and paste information from various sources like Google Analytics and Google Ads. It’s a lot of work, but they're making it happen.

Growth and the CMO's vision

As their company grows, they hire a CMO, who recognizes the need for a more structured approach. She pushes for dashboards and reports, aiming to move beyond the limitations of Excel. Some reports are built on the existing Excel data, while others pull information directly from CRM and advertising platforms.

The data team's impact: Defining metrics

As the company grows, they get to the point where the CMO can no longer single-handedly manage their data needs. In response, they hire a small but dynamic group of data professionals. The new data team has a mission – to distribute self-serve, trusted data that empowers informed decision making.

The team begins by defining metrics and creating composable artifacts. To avoid bottlenecks and to ensure data integrity, they add certified metrics to a central catalog and govern access with roles and permissions. Mission accomplished! Stakeholders have access to single-source-of-truth (SSOT) data for exploration, analysis, and intelligent decision making.

Our two young entrepreneurs now have the data they need to continue to grow their business and make the right decisions at the right time!

The great divide: Data experts vs decision makers

As companies grow, two data-related groups tend to emerge.

• Data Experts (Data Team): These professionals handle the technical aspects of data management, for example, sourcing and architecting the data, and managing access to the data.
• Decision Makers (Business Users): These individuals rely on data, most often in the form of dashboards or reports, to make informed business decisions.

The challenges of traditional BI

Creating effective dashboards and reports requires an in-depth understanding of the underlying data. The data team has the necessary skills to perform these tasks, however, relying on them to provide routine reporting can lead to inefficiency and delays:

• Data Team Overburden: When data team members are dealing with ad-hoc requests to create and update reports, they’re being diverted from the job they’re professionally trained to do – that of managing the company’s data. This reduces their efficiency and hinders their ability to maintain data quality.
• Decision-Making Delays: Decision makers often have to wait for data team members to deliver reports. This bottleneck can lead to delays in crucial business decisions and erode your timely response to changes and opportunities.

Metrics-centric analytics: A new approach

A metrics-centric analytics approach proposes a new type of data artifact, one that acts as a bridge between the data team and business users.

• Effective Data Teams: In a metric-centric infrastructure, the data team can concentrate on managing and defining metrics, instead of report creation.
• Self-Serve Business Users: Metric-centric analytics empower business users to access and analyze data independently, enabling timely decision-making.

Some important terms

Before diving in, let's define a few metric-related business terms:

Key Performance Indicator (KPI)

Key Performance Indicators (KPIs) are used to measure the health and success of a business. KPIs are often monitored in reports and on dashboards. Revenue is an example of a commonly-tracked KPI.

Objectives and Key Results (OKRs)

Objectives and key results (OKRs) are the goals set within a particular time period based on KPIs. For instance, if revenue is a KPI, the OKR might be to increase revenue by a specific percentage by the end of the quarter.

Metrics

Metrics are data artifacts within a system that can be queried. They’re often used to track KPIs, with associated goals, to monitor progress towards OKRs. There are also secondary, supporting metrics that aren’t directly tied to KPIs.

Why metrics are better than traditional BI data models

A metric is a data model with a single, well-described business definition, written in familiar, easy-to-understand business language. A metric measures one thing. You may have many metrics in your system, but each one has its own unique meaning.

Metrics are composable, meaning they can be combined together to create new and richer sets of metrics.

Metrics only expose query operations that don’t alter their meaning. For example, aggregation choice is not exposed because the same data aggregated differently has a different meaning – a “total sales” metric is different from an “average sales” metric  even though they use the same underlying data. However, metrics do expose  capabilities that allow you to query the metric for different contexts, for example, to analyze data by month, region, or product type.

In traditional BI, reports and dashboards tie meaning to the data and expose that meaning through presentation logic. Metrics don’t enforce presentation logic or usage. Instead, metrics tie the meaning to the data directly, but don’t control how you use it. 

What makes a metric?

A metric is composed of values that are associated with a date/time and, often, dimensions that provide context for the values being measured. Behind the scene, metrics also include query rules and configurations.

Let’s take a closer look:

• Values are the quantities being measured. Each metric tracks a single value, for example, revenue, net profit, or lead conversion rate.
• Each value is associated with a date/time (timestamp) that indicates when a value was recorded or to what time it relates to. The combination of values and timestamps defines the core meaning of the metric.
• Most metrics include one or more dimensions. Dimensions provide additional context for the metric and allow data to be broken down for comprehensive analysis.
• Metrics are powered by a set of query rules and configurations that determine how the data is queried and prepared by the system. These rules and settings are defined by the metric’s creator and are not exposed to end users.

The following example shows how the values, timestamp, and dimension components map to a visualization. The values become the data points. The timestamps create the primary time axis, enabling the data to be viewed over time. The dimensions allow for data segmentation for in-depth analysis.

Metric-enabling technologies

There are a few relatively new technologies that support and enable the creation of metrics.

• Metric platforms
• SQL metrics and data warehouse metrics
• Semantic layers

Metric platforms

Metric platforms, typically SaaS offerings, provide a complete data stack optimized for metric storage and querying. These platforms include metric definitions and metadata descriptions. They also enable the collection of historical data, supported by regular refreshes/data imports to increasingly build metric history. This is especially beneficial when tracking data from API sources, like Google Analytics and QuickBooks, where querying current data or short time windows is common practice.

Metric platforms are also great for data stored in spreadsheets, CSVs, or other files. For businesses transitioning from spreadsheet-based data management, these platforms facilitate the initial setup of metrics. Existing spreadsheet update processes can continue, with subsequent imports into the platform to update the metric data.

Level up data-driven decision making

Make metric analysis easy for everyone

Get Started Now

SQL metrics and data warehouse metrics

Metric systems that expose SQL metrics or connect directly to data warehouse metrics are becoming increasingly popular. In these scenarios, some modeling within the warehouse may be required to prepare tables or views to connect to metrics, but the data remains in the data warehouse. Some modeling within the warehouse might be required to prepare tables or views for the metric connection. Once configured, the metric can be linked to the table fields with query options set to translate metric queries into SQL queries against the database tables.

Semantic layers

Semantic layer integrations, for example, dbt™ Semantic Layer and Cube, are also on the rise. These layers enable rich modeling within the data warehouse and support advanced use cases. They add a layer of metadata that metric systems can use to inform their query options. Connecting a metric to a semantic layer typically involves pointing to the fields in the layer from which the metric system can determine how to query them. Some semantic layers have their own built-in metrics layer.

Why metrics?

There are many reasons why metrics are the building blocks of modern self-serve BI. Their inherent qualities make them reliable and dependable while, at the same time, flexible and adaptable.

Metrics are universal

No matter what type of system or combination of systems you use, metrics will always function and behave similarly and consistently. So much so that you can redefine a metric from one type to another with no impact on downstream consumers.

Metrics can be combined using calculations

Metrics can be combined, using formulas, to create new metrics. The operands within the formula are other existing metrics of any type, even other calculated metrics.

In calculated metrics, the system queries the underlying metrics using the same query. Then, the aggregated results are joined and the formula is applied to each row.

The newly-created metric functions exactly like any other metric. It adheres to the same rules, is queried in the same way, and can be used to create additional combined metrics.

Metrics can be combined in visualizations

Combining metrics in visualizations enables you to create the same rich dashboards and reports found in traditional BI applications. Metrics are singular in their meaning but visualizations don't have to be. You can combine multiple metrics in a single visualization by joining and aligning them using a commonly-shared context.

Here's an example of two metrics visualized separately and combined using a shared time context.

This ability to combine metrics, a bit like building with Lego blocks, is an easy and approachable way for end users to create rich, complex visualizations without impacting the underlying data – perfect for self-serve reports and dashboards!

Metrics support single source of truth data

Reliable, trusted data is a key component in informed decision making. Centralized metrics, managed and certified by the data team, and governed by access roles and permissions, make single-source-of-truth (SSOT) data available to everyone. This central repository, often referred to as a catalog, is not only an access point for the company’s metrics, it’s also a library where users can quickly check metric progress and get an overall sense of the business's health and performance.

Metrics make it easy to visualize and report on data

Metrics inspire confidence by making it easy to create visualizations, dashboards, and reports. And, perhaps even more importantly, because metric meanings can’t be altered, users know they can trust the data they’re working with. Even those without an in-depth understanding of the underlying data architecture can confidently use metrics to generate insights.

Metrics enable advanced analytics

The structured nature of metrics supports advanced analytics. For example, time series analysis depends on a consistent time axis with specific granularity. Metrics have a natural time axis and consistent query language, perfect for this type of advanced analysis.

Metrics are queryable and easy to integrate

Metrics can be queried just like any other data. Metric systems often feature their own query language  and many also offer a subset of SQL for accessing the metrics. With this ability, you can use JDBC drivers to easily integrate metric data into other systems, much like you would in a traditional database.

Getting started with metrics

Now that we’ve learned about the advantages of metrics, let’s talk about how to get started:

1. Collaboratively perform a needs analysis:

Initiate discussions between the data team and the decision makers to identify OKRs and KPIs. Look at your existing data to make sure you have everything you need. Before moving on, work together to prioritize the KPIs and OKRs. These details will help the data team create an efficient delivery roadmap. Remember – this should be a collaborative effort where the data team is fully involved and understands what the KPIs are and how they can be satisfied with data.

2. Implement the required data architecture:

Now that you have a KPI roadmap, it’s time for the data team to implement the data architecture to meet its requirements. This is a crucial step where the data team works to ensure there’s a solid foundation for the company’s metrics.

3. Define key metrics:

Adhering to the roadmap, the data team defines the company’s metrics in order of priority. Throughout the process, they ensure each metric is defined in business terms with useful, consistent naming and descriptions. They also include calculations and data sources in the definitions to show where the data is coming from.

4. Set up access controls for self-serve analysis:

To maintain SSOT, trusted data, it’s essential to set up access control for the company’s metrics. Using a set of roles and permissions, end users are able to self-serve certified, approved metrics from a central catalog for analysis, reports, and dashboards.

5. Establish a feedback and improvement loop system:

As time passes, metrics will inevitably need to be added, updated, or removed. Create an internal process where these changes can be requested and prioritized.

Metrics, the self-serve BI artifact for modern analytics

Metrics are the obvious solution to achieving self-serve access to data analytics. As ideal consumption artifacts, they address a decades-long gap in BI. With a consistent yet adaptable structure, they provide safe, familiar access to trusted data while also enabling advanced analysis. Metrics, meticulously and consistently defined, centrally managed and governed, create a vital bridge between data teams and decision makers – A bridge that guarantees end users access to the right data at the right time, a perfect combination for confident, informed decisions.