Predictive Models - Snowflake Setup

There are various ways in which you can send/receive data via Snowflake. These steps cover the preferred method by Resonate, which is to send us your data via Snowflake via Data Sharing. Please get in touch with your Resonate Solution Engineer if you prefer a different option.

Important Notes: These steps use Snowflake's data sharing option (not the cleanroom option). Resonate supports all commercial Snowflake regions — your account does not need to be in a specific region.

Overview

Overall, there are 5 steps you’ll follow to share and receive Resonate Append data via Snowflake:

1. Create one table for the CRM data that will be used for predictions using the Resonate prescribed format in your Snowflake account
2. Create another table for the manifest detailing/defining the CRM data
3. Send the data to Resonate
4. Resonate processes the tables and creates the outputs within SLA timeframe
5. You will receive the following deliverables back from Resonate:
   1. Scored CRM table on Snowflake

   2. A results document (Word) covering methodology, your match rates, decile compositions, and model performance

*Repeat steps 1-5 for each Predictive Model refresh

Step 1: Create the CRM data table in your Snowflake account

• In your Snowflake account, create a table that will be used to hold your CRM data that Resonate will utilize to predict churn or your next best customers
• The table must be created using the Resonate prescribed format as detailed. Please review the associated Data Requirements article for more details.
• If you have the file in CSV format after following the associated data requirements document and wish to make it into a snowflake table, Snowflake supports table creation through CSV. Make sure to select the first line of the CSV as the header. 

Step 2: Create the manifest data table in your Snowflake account

• In your Snowflake account, create a table that will represent the manifest that describes/defines your CRM data so that Resonate can properly predict on it.

• The table must be created using the Resonate prescribed format as detailed. Please review Article 2 (Data Requirements) for more details.

• If you have the file in CSV format and wish to make it into a Snowflake table, Snowflake supports table creation through CSV. Make sure to select the first line of the CSV as the header. As a note, column is a special keyword in Snowflake, so when you are creating the table you will need to put quotes around that name.

Step 3: Send the data to Resonate

• Share both tables with Resonate once the data has been loaded. There are a few different ways in which you could share the data. Our preferred method is as follows:

  • In Snowflake, navigate to the "Private Sharing" option in the left navigation menu.

  • Click on "Share" and select "Create a Direct Share".

  • Select the two data tables you want to share and click Done.

  • Add the Resonate Snowflake account identifier provided by your Solution Engineer in the Add accounts field, then click Create Share.

At this point you have successfully shared the data with us.

Step 4: Resonate will process the data and create an output table for you in the expected SLA timeline.

• Please contact your Resonate Solution Engineer and let them know that the data has been shared.

Step 5: You will receive 2 deliverables from us

• Output Table containing the scored data — your Resonate Customer Engagement Manager will notify you once the output file is shared with you. In Snowflake, navigate to "Private Sharing" on the Snowflake account you initially shared the tables from and you will notice a new table shared by Resonate which contains your scored data.

• A results document (Word) covering methodology, your match rates, decile compositions, and model performance.

The Data Share is now complete.

Questions? Contact your Customer Success Manager or resonatesupport@resonate.com.