Adobe Analytics to Snowflake

[parthpa14]

What is the best possible way to get all the clickstream data directly from adobe analytics to snowflake?

[pradnya_balvir]

Hi @parthpa14 ,

1. Adobe Data Feeds

   • Request raw clickstream data via Adobe’s Data Feeds (available in CSV format on an FTP/S3 location).

   • Data includes hit-level details (one row per hit), ideal for granular analysis.

2. Automated Ingestion Pipeline

   • Use an ETL/ELT tool like Fivetran, Informatica, Matillion, or Airflow to:

     • Ingest files from S3/FTP.

     • Transform and load them into Snowflake.

3. Optional: Adobe Experience Platform (AEP)

   • If using AEP, leverage Real-Time Customer Data Platform (CDP) for streaming data into Snowflake via APIs or event forwarding.

Thanks.

Pradnya

[parthpa14]

Hi Pradnya,

Can you share more information regarding option 3: Adobe Experience Platform (AEP). Especially interested in leveraging the APIs.

Thanks,

Parth

[Jennifer_Dungan]

Yes, seconding what @pradnya_balvir said... we use the Raw Data Feeds to an S3 Bucket, and Airflow to get into Snowflake.

Some notes about the Raw Data feed... its best practice to use as much delay on the feeds as possible, this gives Adobe more of a chance to finish processing the data before sending the files off... the raw data feeds don't have "delta logic" sending everything that wasn't sent in the last file....

Originally our team was losing a lot of data because the original team set up hourly feeds with no delay... this resulting in absolutely no mobile app data, and partial web data; as there was no time left for processing. Even if you are doing a daily feed, I suggest adding a delay so you don't lose the rows of data that come in close to midnight.

[Vinay_Chauhan]

Hi @parthpa14 

The most reliable way to get Adobe Analytics data into Snowflake is by using Adobe Data Feeds.
These give you raw, hit-level data (1 row per server call), which is ideal if you’re looking for full clickstream-level visibility.

Steps for setup -

1. Set up Adobe Data Feeds to export data (usually as .tsv or .csv) to either an S3 bucket or FTP location.

2. Use an ETL tool like Airflow, Fivetran, Matillion, or Informatica to pick up the files and load them into Snowflake.

3. From there, you can handle any transformations or modeling as needed.

Please remember, when scheduling the data feed in Adobe, make sure to add a processing delay (e.g. a few hours for hourly feeds, or at least 4–6 hours for daily feeds).

This gives Adobe enough time to finish processing events, especially mobile app data, which often comes in with some lag. If you don’t add a delay, you risk missing partial or late-arriving data.

If you're already using Adobe Experience Platform, you might also consider using Event Forwarding or Streaming APIs, but for most setups, Data Feeds + ETL remains the most direct and scalable approach.