Documentation

The Blend Data operator is a powerful tool in Self-Serve Data Transformation, allowing you to merge and unify data from multiple sources into a single, cohesive dataset.

How Does It Work?

1. 1‍Define the Output Schema
   Blend Data begins with defining the output schema, which determines the structure of your resulting dataset.
   In most cases, you’ll start with a pre-configured Recipe, where most fields and mappings are already prepared.

2. 2‍Add Sources
   Select the data sources you want to include in the blend.
   These can be tables from your database, datasets from external platforms, or transformation results from previous steps.

Click to open in full screen 👀

3. 3‍Map Fields
   Match fields from each source to the defined schema. This step can be simplified using AI Autofill, which analyzes your data and automatically maps the fields based on output schema.

Click to open in full screen 👀

4. 4‍Apply Custom Formulas
   Enhance the blending process by applying Custom Formulas to your fields: use formulas to calculate new fields, transform data by specific conditions. For detailed guidance on creating and applying formulas, refer to the Custom Formulas Documentation.

5. 5‍Run and Save
   Once your schema, sources, and mappings are configured, click «Run and Save» to execute the blending operation. The result is a unified dataset, structured according to your schema and incorporating all applied formulas and mappings.

Use Case

Blending Facebook, Bing, and Google Ads Data

To illustrate, let’s walk through a real-world example of blending data from multiple advertising platforms: Facebook Ads, Bing Ads, and Google Ads.

Goal

Create a unified dataset with standardized metrics and dimensions.

Steps

1. 1Define the Output Schema: Start with a pre-defined Cross-Channel Recipe, which includes the appropriate output schema.
2. 2Add New Sources: The default recipe starts with Google and Facebook data sources. We just need to add one more, Bing Ads.
3. 3Map Fields: Use AI Autofill to automatically map fields from the sources to the schema.
4. 4Run and Save: Execute the blending operation to generate the final dataset.

The result is a unified, standardized dataset that integrates fields from Facebook, Bing, and Google Ads. This dataset is now ready for cross-channel analysis, offering insights into performance metrics like total spend, impressions, and clicks across platforms.

Conclusion

The Blend Data operator is an intuitive yet powerful tool for unifying datasets from multiple sources. By leveraging recipe, AI-powered field mapping, and custom formulas, it simplifies the process of creating actionable datasets.