Flow Chart

A step that happens again later in a case is drawn as a dashed arrow back to the earlier step, and a step repeated directly is a small loop on top of its box. Layout > Same Name Is One Step (Loops) keeps one box per activity, so rework shows as rework instead of a second copy of the step.

When two steps of a case share a timestamp, their order is a guess: bind Sort Order (a step number) to settle it, otherwise ties sort by activity name. The data table opens with a data-quality check that counts rows left out, events ending before they start, single-event cases and steps sharing a timestamp, so you know how far to trust the log before anyone reads the map.

Yes. A Box Shape field picks the symbol per step: process, decision, start and end terminators, document, data, manual, preparation, subprocess, database and delay. Box Image adds an icon per step from data:image values in your data, Note writes a short annotation under the step, and Link puts a link icon on the box that opens a work instruction or a system page after a click (https and http addresses only).

With a two-level Activity hierarchy, Layout > Group Steps into Subprocesses draws each group as one box with the subprocess symbol and an expand button; the hierarchy also drills down. Search, collapse and expand, step numbers, a background grid, level headers, a legend, pan and zoom, an overview map when zoomed in and boxes you can drag into place (saved with the report) are all free.

That is process mining lite, and it is Premium. Durations shows the median, 90th percentile or mean time on every arrow, measured from the end of one step to the start of the next, never negative; with End Timestamp bound it splits waiting from processing time and draws the lead-time distribution. Bottlenecks colours the arrows from fast to slow and numbers the slowest, SLA flags cases and steps over their target, and Rework shows the rework rate per step.

Conformance checks every case against the most frequent variant, a typed path (A > B > C) or several accepted variants, and marks on-path arrows, skipped and inserted steps. The variant explorer lists every variant with its share, lead time and share month by month. Swimlanes counts handoffs between departments, Compare By puts two periods or segments on one map, Small Multiples draws one map per site, the case list names the slowest cases behind a step or an arrow, and the time-true replay moves every case through the map at its real waiting times.

The event log reads up to 300,000 events, and there is no node cap in the free version. For bigger logs, aggregate the log to From, To, Value and Duration in Power Query and bind the pairs: ten million events become a few hundred rows and the map, loops, labels and average times stay exact. Layout > Keep the Top N Steps merges the rarest steps into one Other steps box with every count exact, and Show Variants Covering (% of Cases) draws only the paths that cover that share of cases. A click on an arrow or variant with more than 1,000 cases filters the other visuals by Case ID.

Directly-follows counts, transition times, variants, rework and conformance are tested against an independent process-mining reference, and percentiles against numpy and scipy. Every time in a tooltip says how it was measured, so "median 19.9 h" means the median of the waits on that arrow, not an average of averages.