How tasks are counted

<h2>The definition</h2><p>One task is one record successfully written to a destination app. A flow that reads a new payment and creates a contact uses one task. A flow that reads the same payment and writes a contact, a card and a chat message uses three, because three records were written.</p><p>Reading is free. Checking a database every three minutes all month costs nothing, no matter how many times we look, because nothing was written anywhere.</p><h2>What counts and what does not</h2><table><thead><tr><th>Event</th><th>Counts as a task?</th></tr></thead><tbody><tr><td>A record written to a destination app</td><td>Yes, one per record</td></tr><tr><td>A lookup that finds an existing record to update</td><td>No</td></tr><tr><td>The update that follows that lookup</td><td>Yes, one</td></tr><tr><td>A run stopped by a filter</td><td>No</td></tr><tr><td>A polling check that finds nothing new</td><td>No</td></tr><tr><td>A test run from the builder</td><td>No, up to fifty per day</td></tr><tr><td>A failed run that wrote nothing</td><td>No</td></tr><tr><td>A backfill of past records</td><td>Yes, one per record written</td></tr></tbody></table><h2>Retries</h2><p>When a write fails for a reason that might clear on its own, such as a rate limit or a provider timeout, Syncline retries three times with a widening gap: one minute, five minutes, then twenty. You are charged once for the record if any attempt succeeds, and nothing at all if they all fail.</p><blockquote><p>Note: a run that fails permanently, such as a rejected field value, is not retried and is not billed. There is no way to lose tasks to a broken mapping, which means you can leave a misbehaving flow on while you debug it.</p></blockquote><h2>Watching the total</h2><ol><li><strong>Open Settings, then Usage.</strong> The bar at the top shows tasks used this billing month against your plan limit.</li><li><strong>Sort the table by tasks.</strong> One flow is usually responsible for most of the number, and it is rarely the flow people expect.</li><li><strong>Set a threshold alert.</strong> Choose a percentage and an email address. Eighty percent gives you enough room to add a filter before anything stops.</li></ol><h3>Plans and limits</h3><table><thead><tr><th>Plan</th><th>Tasks per month</th><th>Check interval</th><th>History kept</th></tr></thead><tbody><tr><td>Free</td><td>500</td><td>15 minutes</td><td>7 days</td></tr><tr><td>Team</td><td>10,000</td><td>3 minutes</td><td>30 days</td></tr><tr><td>Business</td><td>75,000</td><td>1 minute</td><td>90 days</td></tr></tbody></table><h3>What happens at the limit</h3><p>Flows pause rather than dropping records, and everyone with an admin role gets an email. Nothing is lost while you are paused: the records are still sitting in the source app, and a backfill after you upgrade sweeps up to seven days of them. Usage resets on your billing date, not on the first of the month.</p>