Best Practices

Guidelines for building workflows that are reliable, easy to maintain, and performant — from naming conventions to error-handling patterns used in production automations.

Project & Workflow Organisation

• One workflow per logical task (e.g., ExtractInvoiceData, SendApprovalEmail)
• Group related workflows in a single project folder
• Use Invoke Workflow to compose small, reusable workflows rather than building one giant sequence
• Keep workflow files under version control (e.g., Git) alongside your other code

Naming Conventions

Variables & Expressions

• Declare all variables at the top of the workflow — not inline inside loops
• Use strongly-typed variables (String, Int32, Boolean) rather than Object where possible
• Avoid magic strings — assign repeated literals to a variable at the top (e.g., baseUrl = "https://app.example.com")
• Use expressions ({{variableName}}) instead of hardcoding values in properties

Error Handling

• Wrap every UI interaction (Click, TypeInto, GetText) in a Try/Catch block
• Log the exception inside Catch: Log: "Error at step X — " + lastException.Message
• Use Retry (3 attempts, 500 ms delay) for flaky selectors before marking as failed
• Set Continue on Error = True only for truly optional steps — never for critical path actions
• Always close the browser / application in a Finally block to avoid orphaned processes

Selector Stability

• Prefer automationid or name over positional attributes like idx
• Use wildcards (~ contains, ^ starts-with) for dynamic text that changes between runs
• Validate selectors with the Selector Editor → Validate button before saving
• Avoid recording selectors from loading / transitional screens — wait for the page to fully load first

Performance

• Minimise Delay activities — replace fixed delays with WaitForElement or WaitForText
• Open Excel once per workflow using ExcelApplicationContainer — do not open/close inside loops
• Use DataTable + ForEachRow for batch operations instead of row-by-row API calls
• Prefer HttpRequest over browser automation for APIs — it is 10–100× faster
• Run independent branches in Parallel to reduce total execution time

Logging & Observability

• Add a Log at the start and end of each major section: "Starting invoice extraction" / "Invoice extraction complete — 42 rows processed"
• Log input variable values at the top so runs are reproducible from the Output Panel alone
• Use Log(level: Warning) for recoverable issues, Log(level: Error) before rethrowing exceptions
• When connected to Orchestrator, all log messages are captured in the Job history — keep them concise and searchable