We have hit our database size limit in Marketo, and found that we are syncing a number of people who are not opted in to email. We are managing email preferences with custom SFDC picklist fields.
1. Is it possible to sync only records that are marked "subscribed" in at least one of these picklists?
2. If we deleted records that are not "subscribed" from Marketo (but not Salesforce), is there a way to get them to re-sync to Marketo when a field value changes?
Any tips or ideas welcome! Thank you!
You can have Marketo support set up a sync filter that will block leads/contacts from syncing. I'd recommend doing a couple of things here. First, create a "Do Not Sync to Marketo" boolean field in SFDC on leads and contacts. Then, have a SFDC workflow that checks/unchecks this when your criteria is met. This will help when troubleshooting issues and help you in the future if you decide there is additional criteria to block other records from syncing. When you're ready to sync the records back to Marketo, uncheck the boolean field either manually or by workflow and during the next sync, that record will be pushed to Marketo.
One issue that you will most likely run into are duplicates. Since you'll have leads in SFDC that are not in Marketo, if they're created in Marketo (maybe form fills, integrations, list uploads, etc), Marketo won't know that records already exist in SFDC and will create them as a new record in Marketo and sync it to SFDC, creating a duplicate. You'll need to set up a merging process to remove the duplicates, remembering to select the Marketo record as the master and ensuring the winning record does not have the "Do Not Sync to Marketo" field checked.
Let me know if you have any other questions.
Agree with the Kyle's approach here to block the records syncing from SFDC to marketo. At the same time, you mentioned, you have hit the marketo DB limit so assuming the problem is more wider then just blocking the unsubscribed records syncing from SFDC to Marketo. To do a bit more data cleanup I marketo, you should be looking at the following as well: