1. Confirm you're on the Scale plan
Browser checks are a Scale-plan feature, capped at 10 per account -- the create form on the Monitoring page shows an upgrade prompt in its place on every other plan.
2. Build the step list
Add steps from a fixed set of seven actions: goto (navigate to a URL), click, fill, waitForSelector, assertText, assertVisible, and screenshot. Each step is a plain selector-and-value row, not a script -- there is no scripting escape hatch, so a flow that needs a conditional or a loop can't be expressed here.
3. Save and watch the first run
Browser checks run from one region (iad) no faster than every 5 minutes. The first step that fails ends the run and names itself in the down reason; a screenshot is kept for every run, and a full failure trace is kept when a run fails.
4. Read a failed run
Open the check's result to see which step failed and the screenshot of the page at that point -- useful for the common failure mode a plain HTTP check can't see at all: the request succeeded, but the page rendered blank or the button never became clickable.
Go deeper
The full reference lives in the docs: Browser checks documentation. Error codes named above are each explained in the error-code reference.