Sync data from
Databricks to Pinterest Ads

Model your Databricks data using any of these methods

• dbt Model Selector

  

  Sync directly with your dbt models saved in a git.

• SQL Editor

  

  Create and Edit SQL from your browser. Hightouch supports SQL native to Databricks.

• Table Selector

  

  Select available tables and sheets from Databricks and sync using existing views without having to write SQL.

• Customer Studio

  

  For less technical users, pass traits and audiences from Databricks using our visual segmentation builder.

Does this integration support in-warehouse planning?

Yes, if you integerate Databricks and Pinterest Ads using Hightouch, in-warehouse planning is supported.

Great, but what is in-warehouse planning?

Between every sync, Hightouch notices any and all changes in your data model. This allows you to only send updated results to your destination (in this case Pinterest Ads). With the baseline setup, Hightouch picks out only the rows that need to be synced by querying every row in your data model before diffing using Hightouch’s infrastructure.

The issue here is this can be slow for large models.

Warehouse Planning allows Hightouch to do this diff directly in your warehouse. Read more on how this works here.

Why is it valuable to sync Databricks data to Pinterest Ads?

Thanks to Databricks, it's easier than ever to access your customer data, run complex queries, and segment your customers/users into various categories or audiences.

However, to target users and send conversion data to Pinterest Ads, you need access to all of the unique behavioral data (e.g., last-login date, items in cart, pages viewed, etc.) and core business metrics like lifetime value, workspaces, subscriptions, annual recurring revenue, etc. that lives in Databricks. You then need to be able to package this data to send to Pinterest Ads in the format expected. Plus, you want to do this in a real-time and automated way.

Pinterest Ads is only as good as the data you give them, and if you truly want to optimize your ad spend, increase your match rates, and drive conversions, you need to provide custom data from Databricks within your warehouse.

Maybe you want to retarget users who abandoned their shopping cart in the last seven days, or upload a list of high-value customers to identify potential lookalike audiences, or perhaps you want to upload offline conversion events to reduce your customer acquisition costs and increase your return on ad spend.

Why should you use reverse ETL to connect Databricks and Pinterest Ads data?

In the past, uploading customer data to Pinterest Ads meant hopping back and forth between your various SaaS applications or asking your data team for CSV files. Neither of these options is preferable because marketing teams want to self-serve, and data teams don't enjoy constantly fulfilling one-off marketing requests.

Even worse, if you truly want to optimize your advertising campaigns, you need to be uploading fresh data consistently. Non-fresh data can be expensive for ads. If you're using CSVs to define who to exclude from paid ads and you're uploading that data weekly, that's potentially one week of irrelevant ads. As a workaround, engineering teams will integrate directly with the Pinterest Ads API, and build and maintain custom in-house pipelines. The problem is that a single API change can break everything, and data engineers don't want to spend their time building and maintaining pipelines.

With Hightouch, you can leverage the existing data models and customer segments your engineering team has defined in your warehouse and sync that data directly to your ad platforms in real time. You can schedule your data syncs to run automatically, on a set cadence, or even for the exact duration of your marketing campaign. Hightouch lets your data teams establish the guardrails for your marketers to self-serve and build custom audiences through a drag-and-drop interface.