Here are a few things you can do in Recipe, but not in Dataflow.
Rename and Save as
Yes, you can download JSON in dataflow and upload it to a newly created dataflow with a new name.
Add Description in each node
Unless you have the EADashboardHelper extension, you can't add descriptions in Dataflow
Join (left, right, outer, inner, cross)
This has been the power of recipes from the beginning
Preview data
Some users would like to see the data after transformation without the need to run it
Column profile
Analyze data at a glance (with a sampling of data)
Direct Input data
If only the objects are not synced to the CRM Analytics platform, this useful to get real-time data from Salesforce
Extract data
Able to write data to Output Connection and CSV
Easier to aggregate, bucket, trim, substring, replace, split, concatenate
You still can do those transformations in dataflow, but just easier in the recipe
Rename field label and API name
You can create a new field using computeExpression in dataflow, but cannot rename the existing field
New machine learning transformations not in dataflow
- Detect Sentiment
- Time Series Forecasting
- Time Series Forecasting
- Predict Missing Values
- Cluster
- Cluster
A few other things worth mentioning
- Prefix in Dataflow = 02K, recipe: 05v
- Unable to run recipe when there is terminating node without write dataset
- Edit recipe in JSON is complicated compared to dataflow
Is anything else not listed above or incorrect? Comment below so I can add them, thank you!
Reference: