Read the run history

<h2>Where it lives</h2><p>Open any flow and the second tab is History. It lists every run newest first, one row per record the flow touched, with the retention window your plan allows: seven days on free, thirty on Team, ninety on Business.</p><p>This is the first place to look for any question starting with why. It answers where a record went, which value broke a mapping, whether a flow ran at all last Tuesday, and how long the destination app took to reply.</p><h2>Anatomy of a run row</h2><table><thead><tr><th>Column</th><th>What it holds</th></tr></thead><tbody><tr><td>Started</td><td>The time the run began, in your own timezone</td></tr><tr><td>Status</td><td>Delivered, skipped, failed or retrying</td></tr><tr><td>Source record</td><td>The ID and title of what the trigger read</td></tr><tr><td>Destination record</td><td>The ID the destination returned, as a link where the app allows one</td></tr><tr><td>Duration</td><td>Total time, split between our processing and the app's response</td></tr><tr><td>Tasks</td><td>How many tasks this run consumed, usually zero or one</td></tr></tbody></table><h2>Opening a single run</h2><p>Click a row and you get three panels. <em>Read</em> shows the raw record as the source app returned it. <em>Mapped</em> shows what each destination field was set to after casting and defaults were applied. <em>Response</em> shows what the destination app said back, including its own error text when something went wrong.</p><p>Comparing Read against Mapped answers most questions in about ten seconds. A value that looks right on the left and empty on the right means a mapping problem. A value wrong on both sides means the source record itself is wrong, and no amount of flow editing will fix that.</p><h2>Filtering and replaying</h2><ol><li><strong>Set the status filter to Failed.</strong> On a busy flow this is the difference between four rows and four thousand.</li><li><strong>Narrow the date range</strong> to the window where the behaviour changed.</li><li><strong>Search by record ID</strong> when a colleague asks about one specific customer or card.</li><li><strong>Open one failure and read the Response panel</strong> before you look at anything else. The destination app usually names the exact problem.</li><li><strong>Fix the flow, then click Replay</strong> on the failed rows. Replay reuses the record we already read, so nothing is fetched twice and no duplicate is created.</li></ol><blockquote><p>Note: replay uses the current version of the flow, not the version that ran originally. That is what makes it useful for testing a fix, and it also means a replay can produce a different result from the row you are looking at.</p></blockquote><h3>Exporting before the window closes</h3><p>Run history is deleted once it passes your plan's retention window, and there is no archive behind it. If a run matters for an audit, use <em>Export CSV</em> from the History tab while it is still there. The export carries every column above plus the mapped values, and a viewer role gets the file with values masked.</p>