Hi Aaron W., I may be delayed in replying, but here are two mistakes I made and the solutions that worked.
Mistake 1: Defaulting to SQL for everything
Early on, I used SQL for all my transformation logic because that’s what my team recommended. For a file containing customers, locations, sales orders, and exceptions, that meant one query per object type, repeated CSV parsing, and extra steps that all consumed tasks. Then I realized Workato has a Python connector. With an output schema, one step splits the file into a clean array per object, ready for foreach actions. On an hourly recipe, that saved 8 tasks per run, roughly 70,000 tasks a year (8 × 24 × 365).
Lesson: SQL is great for set-based data work, but for complex parsing, check whether a code step can do it in fewer steps.
Mistake 2: Too many API calls for one outcome
Closing tickets needed three kinds of requests: fetch records with a filter, check each record’s description, then close each match. That meant lots of steps and tasks, and the app had no batch-close option. I built a custom connector with the Connector SDK that wraps fetch, match, and close into one action, and added batch closing. A multi-step loop became a single step.
Lesson: if a recipe keeps repeating the same chain of requests, wrap it in an SDK action.
