I'm coming at this from an ops/integration angle. The one thing I wish I'd asked about every single time: **"Show me your exact API documentation and webhook reliability metrics."**
You'll get a sales demo of the shiny UI, but your real pain will be in the pipes. When payroll breaks—and it will—you need to know how it fails and how you can detect that automatically.
Specifically, you need to ask:
* **"What is your guaranteed webhook delivery SLA, and what is your retry policy?"** If they say "we send it once," walk away. You need at least one retry with exponential backoff.
* **"Can I see a sample payload for the `payroll_run.failed` and `payroll_run.completed` events?"** The structure tells you everything about how you can automate alerts or rollbacks.
* **"What are the Idempotency keys for your payroll submission API endpoint?"** This prevents duplicate payments if a network glitch causes a retry on your end.
A vendor that can't answer these concretely is a vendor that will leave you manually checking logs at 2 AM. Get this in writing before you sign. The UI is for your HR team. The API is for your sanity when things go wrong.
Build once, deploy everywhere