Creating custom data source metrics - First steps and best practices

Want to create metrics but not sure where to start? This article focuses on custom metrics and includes some introductory information and data design suggestions to help you on your journey.

Note: If you're interested in adding instant metrics, you can learn more here.

Let’s begin at the beginning .... What is a metric? Metrics help you determine the health of your business. A metric depicts a single numerical value that’s used to track the status of a business process. Some examples of metrics are ‘Average Revenue per Account’, ‘Customer Lifetime Value’, and ‘Website Sessions’.

When you create a custom metric, you add dimensions, so you can see your performance from different perspectives. Dimensions provide context for the numerical value in your metric and are used to break down, filter, or group your data. Every metric includes at least one time dimension. Some examples of dimensions are country and product type.

Why PowerMetrics?

With PowerMetrics, you can:

  • Understand your data and its impact on your business, regardless of your level of experience. PowerMetrics is for everyone – from beginner to data analyst. Start with curated instant metrics, available for our most popular services. If you’re interested, step up your data skills with our custom options – write queries and get custom data in the query builder, manipulate your data by applying formulas in the modeller, and combine data from different sources with calculated metrics or merged data sources.
  • Keep track of your performance over time and compare current to historical data. We retain your historical data by storing the latest records every time your data is refreshed, gradually building up your data history.
  • Gain insights into your data through interactive exploration. Dig into your data by adding one or more metrics to Explorer. Try out different visualization types, apply various date ranges, and choose unique ways to segment and filter your data. It’s quick and easy and you can see the impact of your choices as you experiment and explore.

Before building a custom metric

Ask yourself:

  • What business questions do you want to answer?

Having a clear vision of the answers you’re looking for will drive your preliminary data preparation and help you make the right choices when building and exploring your metrics.

No matter what metric you're interested in, make sure you select a data source and dimensions (columns) that include the data you’ll need.

Need some help deciding?

The sky’s the limit when it comes to creating metrics in Klipfolio®. Looking for some inspiration? Take a look at MetricHQ, the world’s largest online resource for metrics.

Once you know the questions you want to answer, then:

  • Consider your data source.

Think about the question you want answered. Does your data source contain the right information? Does it include the numerical values you want to track or records that can be counted? Does it include all the dimensions you want to use in your metric, so you can see your data from several perspectives?

Take a few minutes to identify the columns from your data source that you’ll use for the metric value, the dimensions to segment the data by, and the time dimension. These are the columns from the data source that are stored in your metrics at each refresh.

Consider how you could refine your raw data to optimize it for building metrics. Before modelling your data source, take a look at the data design tips described below.

Data design tips

Here are a few data design best practices:

  • Time and history are key to the value of using PowerMetrics. To fully appreciate this feature, when creating metrics, choose a data source that includes historical data.
  • Using more granular data, like transactions and hourly data (versus data that’s summarized over longer periods, like by week or by month) enables you to leverage how metric values are automatically aggregated as you switch between time periods.
  • When modelling a data source:
  • Choose a rich data source that includes columns that can be segmented in multiple ways. This ensures you'll have lots of options when exploring your metric later.
  • Not all data columns make optimal dimensions. Only include the columns that represent the dimensions you want to see in your metric.
  • Rename columns (optional, depending on the metric's purpose and proposed audience).
  • Don’t “over-prepare” your data. For example, unnecessary grouping or aggregation of your modelled data prevents you from accessing a metric's flexible ability to display varying levels of granularity.
  • Use formulas to manipulate and optimize your data. Using a formula to group your data into categories (or buckets) can bring simplicity and clarity to your metric. For example, grouping large sets of individual numbers into ranges of numbers will simplify your presentation. A clearer message is a more powerful message.
  • Transform multiple columns into smart, combined dimensions, for example, concatenate a first name column and a last name column into a name column.

Don't be afraid to start over!

Great custom metrics aren’t necessarily built on the first try. Creating meaningful metrics is often an iterative process. The creation process is fast and easy so, if you create and explore a metric and it’s not what you want it to be, don’t be afraid to delete it. Return to your data source and either edit it further or choose a new data source before creating your next version. If you build a custom metric and a dimension isn’t useful, then remove it. Less is usually more and there’s almost always a different dimension you can add that will have greater value.

The only exception to deleting a metric and starting over is if it includes stored history that’s not available anywhere else.

Learn more

 

Have more questions? Submit a request