Your path to data enrichment
One significant advantage of the Append process for Resonate Embeddings is its quick and simple workflow. The end-to-end process is laid out below:
1. Establish essentials in a project kickoff
Resonate will help you determine the three following essentials:
A. Set your Embeddings Reset cadence.
- Monthly refresh - Resonate includes monthly refresh of targetable customer IDs—not to exceed purchased records. This ensures you maintain the most recent rolling digital footprint.
- Embeddings reset: You will determine if you want to retune/retrain your models quarterly or annually (we recommend quarterly). Learn more about our Embeddings Refresh schedule.
B. Confirm the ID type and quantity you will provide.
We require a Customer ID and at least one of the following ID types:
- Hashed Emails(HEM): MD5 (preferred), SHA-128, SHA-256
- Mobile Ad IDs (MAID)
- Postal Address and/or ZIP11
- IP Addresses
You'll also determine the number of records in your file needed to achieve maximum lift.
C. Determine data location.
Resonate prefers accepting your data via Snowflake, but we do offer flexibility in secure file sending and receiving to and from your platform of choice:
- Snowflake
- Box
- AWS
- SFTP
2. Pre-work: Preparation and Validation
We have a couple of pre-work steps to take to maximize the accuracy and speed up your delivery.
- Data Preparation: Format your file for delivery according to our file formatting guidelines, we've provided a Hashing Script and best practices to support you.
- Data Validation: This is an optional step, but one we recommend. Resonate will process a smaller representative file of 5-10K records for data attribute alignment, to estimate match rate and ensure smoothest and fastest delivery—You can learn more about it here.
3. Collection and ID matching
Once we've stepped through the pre-work and fully aligned, we're smooth sailing for the rest of the process. We'll collect your files, and begin with a direct match to our database.
4. Append
Within 10-15 days, Resonate will provide the appended Embeddings Vector in the same location you originally provided your file.
Please note, we will refresh on your targetable IDs monthly—but will not exceed the number of purchased records. We will do a full Embeddings reset either quarterly or annually based on what you determined in the kickoff stage as outlined in step 1.
5. Activate in workflow
Embeddings work nicely with most ML algo’s. We’ve prepared some sample code to shepherd your projects to completion using Resonate Embeddings and familiar algorithms. Find our library here: GitHub.
Comments
0 comments
Please sign in to leave a comment.