Question 1

What can I write into Microsoft Ads from Databricks?

Accepted Answer

Custom audiences, where each row the model returns becomes a member, and offline conversion events for conversions that completed away from the click. Audiences are where most teams start, and conversions matter when the sale finishes in a CRM or billing system rather than on a landing page, which is exactly the data already modelled in the lakehouse.

Question 2

Does the audience sync use Delta Change Data Feed?

Accepted Answer

No, and it is worth being precise about this since Change Data Feed exists on the Databricks side. Incremental reads use tracking columns instead: Polytomic records the prior position and adds bounded predicates so later runs read forward from a high water mark. Use a monotonic column such as an updated timestamp. The consequence is that an update which does not move the tracking value, and a hard delete, cannot be discovered this way.

Question 3

Does a raw email address leave the lakehouse?

Accepted Answer

Not in readable form. Polytomic hashes the personal data before transmission rather than handing Microsoft raw values to hash on arrival. Plaintext columns in the model are fine because they are hashed on the way out, and pre hashed columns generally pass through unchanged. Note that Microsoft audiences match on email only, not phone, so plan the model around the addresses you actually hold.

Question 4

Can a suppression list shrink on its own?

Accepted Answer

Yes, if you use mirror mode. Add mode only ever adds and remove mode only removes, but mirror reconciles the audience against the model on every run, adding and removing as the model changes. Microsoft Ads supports it, which not every platform here does, so a list of people you have stopped paying to reach genuinely contracts when somebody becomes worth targeting again rather than accumulating indefinitely.

Question 5

Why build the audience in the lakehouse?

Accepted Answer

Because the definition worth targeting usually depends on a model rather than a filter. Best customers, likely repeat buyers, people who should be suppressed, all of it comes from product, billing and CRM data combined and often scored. Ad platforms cannot express that, and a manual upload freezes it on the day it was made. Reverse ETL keeps the platform in step with whatever the model currently says.

Sync Databricks to Microsoft Ads (Bing Ads)

Sync Databricks to Microsoft Ads (Bing Ads) with Polytomic

Mirror keeps it accurate

Audiences and conversions

Hashed before it leaves

Or sync Microsoft Ads (Bing Ads) to Databricks

Trusted by businesses worldwide

Step-by-step integration guide

Connect Databricks in Polytomic

Connect Microsoft Ads (Bing Ads) in Polytomic

Create a sync between Databricks and Microsoft Ads (Bing Ads)

Map fields and enable the sync

Frequently asked questions

The sync and ETL platform that does it all

Powerful data sync

Advanced analytics

Enterprise security

Easy setup

Ready to connect Databricks with Microsoft Ads (Bing Ads)?