Using a custom metadata type to externalize the routing rules is a best practice for complex, variable logic. It makes the approval process itself simpler and more stable, while allowing admins to easily update approver mappings (e.g., when a regional VP changes) without modifying the process definition. This approach is more scalable and maintainable than hardcoding logic in criteria or creating numerous separate processes.

The most common performance killer in Flows, especially bulk ones, is inefficient data operations. ‘Get Records’ elements, particularly non-selective queries or those placed inside loops that cause the query to execute repeatedly, are the primary culprits for slowdowns. Optimizing these queries and the Flow’s bulkification is the first and most critical step before considering architectural changes.

A Record-Triggered Flow with a Scheduled Path is the modern, declarative standard for time-based actions that require complex logic or data from multiple objects. It replaces Time-Based Workflow Rules (which are less flexible) and the older Process Builder/Flow combination. The Send Email action within Flow is more powerful and integrated than a standard Email Alert for dynamic content, allowing direct access to all Flow variables and data.

A before-save Record-Triggered Flow is the optimal declarative choice for this complex validation. Validation Rules cannot directly perform a lookup or count on child records in their formula. While a Roll-Up Summary field with a Validation Rule would work, it requires an extra field and doesn’t update in real-time during the same transaction before the save. A before-save Flow can query the related records immediately and add a field error, providing the required real-time, transactional prevention.

Process Builders, like other metadata, have unique API names. The ‘Duplicate Rule’ error on activation typically indicates a naming conflict, most commonly because an inactive version of the process (or one with the same name from a different package) already exists in the target org. The first step is to navigate to Setup -> Process Automation -> Process Builder in production, view all processes, and look for any inactive process with a similar name.

Every running instance of a Flow has a unique Interview GUID accessible via the `{!$Flow.InterviewId}` global variable. The most direct method is to assign this value to a Flow variable and then map that variable to a custom field (e.g., ‘Created by Flow Id’) on the Task record when using the ‘Create Records’ element. This provides a clear, auditable link back to the exact Flow interview.

Using a declarative Roll-Up Summary field (using a COUNT or MAX/MIN of a formula field on the child) is the most efficient for this real-time, criteria-based rollup. The calculation is handled by the platform efficiently. A separate automation (like a Record-Triggered Flow on Project) can then watch for changes to that roll-up field and update the ‘At Risk’ checkbox. This is preferable to a child-triggered process that queries and updates the parent, which can be less performant and harder to bulkify.

The critical anti-pattern is performing a ‘Get Records’ (a SOQL query) inside a loop. This causes the query to execute once for each record in the triggering batch, leading to excessive query rows and likely governor limit exceptions. The proper bulkification technique is to move the ‘Get Records’ outside the loop. You would first collect all relevant Account IDs into a collection, then perform a single query to get ALL related Contacts for all triggered Accounts, storing them in a variable. Inside the loop, you then filter or find the Contacts relevant to the current Account from that collection.

Screen Flows are the primary declarative tool for building user-guided processes within the Lightning Experience UI. They can be surfaced as Actions on object record pages, in the Global Actions menu, or in Utility Bars. This is the perfect use case: a button (action) that starts an interactive Flow, which can perform logic, show screens, create records, and end by redirecting the user to a specific record.

Using a custom metadata type to externalize the routing rules is a best practice for complex, variable logic. It makes the approval process itself simpler and more stable, while allowing admins to easily update approver mappings (e.g., when a regional VP changes) without modifying the process definition. This approach is more scalable and maintainable than hardcoding logic in criteria or creating numerous separate processes.0