Skip to main content

​
Data tracing

Data tracing lets you follow a single row of data as it moves through your flow. Pick a row, and Parabola lights up every related row in the connected steps — the source rows that fed into it, and the rows it turned into downstream. It’s the fastest way to answer the question every flow-builder eventually asks: “Where did this row come from, and what happened to it along the way?” Instead of opening each step and eyeballing the data, you trace the row once and see its whole journey across the canvas at a glance.

​
When to use it

Reach for data tracing whenever you need to understand or debug a specific row:
  • A value looks wrong in your output. Trace it back to the source rows that produced it to see where the number went sideways.
  • A row you expected is missing. Trace a related row to see where it got filtered, joined away, or transformed out.
  • You’re learning a flow someone else built. Trace a row end-to-end to see how the data actually flows, step by step.
  • A chart point looks off. Trace a mark in a visualization back to the exact rows behind it.

​
Start a trace

You can start a trace from anywhere you can see a row:
  1. From a step on the canvas — Click the search icon on any step to show data, then click on any row to start tracing.
  2. By clicking a visualization — Click a bar or point directly on a visualization to trace that data.
  3. From a step’s results — Open a step’s full results, double click a cell, and choose Trace this row.
That’s it — the trace starts immediately and the canvas updates to show you the row’s lineage.

​
What you’ll see

Once a trace is active, the canvas shifts into trace mode:
  • A toast appears at the bottom of the canvas: “Row tracing: from your selected row, we’re showing all the related rows in connected steps.” It stays up for the whole trace, with a Stop tracing button.
  • Your selected row is highlighted in the step you started from, which shows Tracing 1 row in its footer.
  • Related rows light up in every connected step — the upstream rows that produced your row, and the downstream rows it became. Each of those steps shows a count like 3 related rows in its footer.
  • Steps your row never reaches are dimmed, so the path your data actually takes stands out clearly.
By default, each connected step shows only the related rows so you can focus on the lineage. To see a related row in the context of the full table, use the (show all) toggle in that step’s footer; switch back with (show trace).

​
Follow the trace further

A trace isn’t a dead end — you can keep pulling the thread. Click any highlighted row in a connected step to re-root the trace on that row. Parabola re-traces from your new pick, following its upstream and downstream lineage. This is how you walk a row backward toward its source, or forward toward your destinations, until you find exactly where things change.

​
What you can trace

Data tracing follows your row across the connected steps in your flow. Steps that are created by the Prowork agent, integrations, and agentic visualizaitons can all participate in tracing. Legacy visualizations and legacy transform steps cannot show tracing.

​
Stop a trace

To exit trace mode, do any of the following:
  • Click Stop tracing in the banner.
  • Press Esc.
  • Click the Stop tracing button on the step you started the trace on.
Trace mode is temporary — it also clears on its own when you leave the flow.

​
Troubleshooting

A step shows “Trace unsuccessful.” This means Parabola couldn’t connect your row to rows in that step — the lineage breaks at that point. It’s often a useful signal in itself: it points to exactly where a row’s connection to its source is lost. “No related rows” in a step. Your traced row genuinely doesn’t have related rows in that step — for example, it was filtered out before reaching it. That’s expected, and itself a clue about what happened to the row. “Couldn’t trace this row. Please try again.” The trace was too large to complete — usually because a step (like a large join) fans the row out into an enormous number of related rows. Try tracing a more specific row, or trace from a step further along in the flow.
Last modified on June 26, 2026