Currently, the custom groupings are used within Attribution models and Budget optimiser. However, we are also working on integrating them into Reporting solutions to create a unified experience of working with the custom groupings across Alvie.
What is a custom grouping?
'Custom groupings' is a concept used within Alvie to categorize large datasets into smaller, logical groups. This helps ensure that the data becomes actionable and usable.
You can think of these groups as a categorization of the data based on other columns in the dataset.
If you're familiar with Google Analytics, you might have seen or created a 'Custom channel grouping'.
If you're more familiar with SQL, you can compare it to a CASE statement.
Both of those approaches are used to categorize the records in a large dataset into smaller groups that are easier to make decisions based on.
For example, when working with marketing activities, it is often beneficial to group data into:
different channels. For example: 'Paid search' or 'Paid social'
or different markets. For example: 'Denmark' or 'Sweden'
This enables analysis and helps uncover insights that are more closely related to the business objectives.
Within Alvie, you will often be asked to provide a Channel grouping or a Market grouping for that exact purpose. This is particularly important when setting up an Attribution model or Budget optimiser, as a custom grouping is essential for the models to understand how to allocate conversions or budget effectively.
We recommend creating new custom groupings only when necessary. In most cases, a single Channel grouping or Market grouping should be sufficient to cover the use cases for a company. However, there may be instances where additional groupings are required, and that is completely acceptable.
Why do we need custom groupings?
The purpose of custom groupings is to categorize data into logical buckets across platforms. This allows Alvie to seamlessly integrate data from different platforms, enabling our powerful cross-platform models and reports to function without relying solely on potentially flawed tracking and tagging.
The visualization below provides a clear illustration of the necessity of custom groupings:
