Presenting a Finding
The discovered map is the least persuasive output. What to show instead, in what order, and to whom first.
Procedure
Process mining findings are frequently presented as a map, which is the format least likely to produce a decision.
Why the map fails as a deliverable
Nobody in an operational meeting can read four hundred nodes.
A filtered map invites the question of what was removed, which then dominates the discussion.
It shows structure, not cost, and the decision is about cost.
It looks like the work, which encourages presenting it as the result.
The order that works
The number first. "Cases with a rejection take 19 days against 6, and that is 22 percent of cases."
Then what causes it, in one sentence.
Then the evidence, briefly, with three real case numbers anyone can check.
Then the proposed change and who would own it.
Then what would be measured afterwards.
The map, if at all, as an appendix for orientation.
Who sees it first
The people who handle those cases, before anyone else.
They will correct it, which is the point, and they will identify what the log did not know.
They will also tell you it is already known, which is frequently true and changes the framing from discovery to evidence.
Presenting upward before validating downward is how findings get rejected in the room, publicly.
Language that survives
Name the specific step and the specific delay.
Give both figures and the difference.
State the boundary and the period.
Say what you excluded.
Say what you do not know, which pre-empts the objection and makes the rest more credible.
What undermines a finding
A percentage with no denominator.
An average with no spread.
A comparison with no normalisation.
A number that surprises the operation and was not checked with them.
Attribution to individuals, which converts an operational discussion into a personnel one and ends the cooperation.
A precise figure resting on a defect, which discredits everything the programme produces afterwards.
The three case numbers
The cheapest credibility device available.
Pick three cases exemplifying the finding.
Include the references.
Anyone can look them up, and someone will.
A finding that survives three people checking three cases is accepted; one presented only in aggregate is argued about.
Closing
One page.
Sent, not only presented, because the decision is frequently made afterwards by someone who was not in the room.
With the owner and the date on it, so it becomes an action rather than an observation.
Presenting to people who run the process
The audience that determines whether anything happens, and the one most often shown the wrong thing.
Do not lead with the discovered model. A spaghetti diagram is read as a criticism of the people in it.
Lead with one number they care about: the waiting proportion, the rework rate, the lead time spread.
Show three variants in plain language, not a graph.
Say what the analysis cannot see, which builds more credibility than any finding.
Ask what is missing, and act on the answer, because they will identify the log defect that would otherwise have surfaced during the challenge.