Consulting & AdvisoryFree
Current State Process Map From Three Sources
What the SOP says, what the people who run it describe, and what the system actually logged, side by side on every step.
River reads the procedures, the interview transcripts and the system exports as three accounts of one process, and builds a step register carrying all three. Where they disagree, the divergence is classified, because a stale document and a compensating workaround look identical in a register and need opposite recommendations. Out comes the current state narrative, the register with the documented version beside the observed one, and an as-is map drawn only from what the evidence actually supports.
Page one has the four sources argument already. Interviews for what people say they do, observation for what they actually do, documents for what the manual says, event logs for what the system recorded, and each source correcting the others. It is right, and it stops there. A register with three columns and no verdict tells a client their SOPs are out of date, which they already know. What decides the recommendation is which kind of out of date it is.
Built for the management consultant, the operations reviewer and the transformation lead who has to produce an as-is anybody will sign. It works off the material collected through the indexed data room and the themes counted in the register built from the interviews. What the process turns out to be feeds the scoring in a diagnostic with observable anchors, and every divergence with a number against it becomes evidence inside the findings and recommendations set.
Regulators already assume the procedure is wrong
Two regulators have written the divergence problem into law, which is the best evidence that it is normal rather than a scandal. 29 CFR 1910.119 requires operating procedures to be reviewed as often as necessary so that they reflect current operating practice, and then requires the employer to certify annually that those procedures are current and accurate. An annual certification exists because documents drift between certifications, predictably, without anybody lying to anybody.
The FDA goes further and hands the register its shape. 21 CFR 211.100 is titled written procedures, deviations. It requires procedures to be followed and documented at the time of performance, then requires that any deviation from them be recorded and justified. Recorded and justified, not eliminated and not hidden. A deviation is a row with a reason attached, which is precisely what a current state register owes a client on every step where the document and the floor disagree.
The system export is the one account nobody can talk out of its position. It carries timestamps, so a divergence gets a start date: the inspection that moved to the following morning appears in the data from the week the depot lost its evening shift. It carries counts, so the branch the procedure disposes of in one sentence turns out to be a third of all recharges. Interviews explain a divergence and give you the reason. The log dates it and sizes it.
How it works
Send all three
The procedures, the interview transcripts, and whatever the system will export about this process.
Bound the process
Where it starts, where it ends, and which of the sites are in scope.
Read the register
Every step three ways, with each divergence classified and the real volumes attached.
Confirm with the floor
The register goes back to the people who run the process, not only to its owner.
What you get
- One row per step carrying the documented version, the described version and the logged version
- Every divergence classified, because a stale document and a workaround need opposite recommendations
- The start date of each divergence, taken from when the system data changed
- Real volume on every branch, so a documented exception carrying a third of cases is named
- Shadow steps that appear in no document, with the spreadsheet and the person holding them
- Provenance per step: procedure version, interview reference, and the export field behind the number
Common questions
What does it need from me?
The procedures with their version and approval date, the interview transcripts, and whatever the system will export about this process. Then the boundary: where the process starts, where it ends, and which sites are in scope. Everything else is inferred from the material rather than asked for twice.
Why classify a divergence instead of just listing it?
Because the class is the recommendation. A stale document gets rewritten and the floor is left alone. A workaround compensating for a broken upstream system stays until that system is fixed. Only a skipped step with a control attached belongs in a risk register. One list produces one wrong recommendation for everything.
How do you put a date on a divergence?
From the export. A step that moved shows a change in its timestamps, and that change lands in a specific week. Once a divergence starts the week a release shipped or a shift was cut, the conversation stops being about discipline and starts being about a decision somebody made.
What is a shadow step?
Real work that appears in no document, usually held in a spreadsheet by one person who has never been asked about it. Shadow steps get drawn on the map and attributed. Half of them are the strongest control in the process and get spread; the other half are single points of failure.
Is this the same as process mining?
Mining tells you what the system recorded, which is one of the three accounts here and the only one that cannot be argued with. It cannot see a step that never touches software, and most of the interesting ones do not. The export sizes and dates a divergence; the interviews explain it. A step nothing records at all needs somebody standing there with a stopwatch.
The procedure calls something an exception. Is it?
Check the volume before you believe it. Documented exception paths routinely carry a third of the cases in their branch, because the procedure was written before the exception became routine. Every path on the map carries its real share, so a redesign cannot spend its effort on the comfortable majority by accident.
Where does this sit in the engagement?
After the interviews and before anything gets designed. The register is what the future state is argued against, and the divergences with volumes on them are the raw material for a scored assessment. Where the process was mapped live in a room, the workshop record is the other half. Which divergences are worth measuring at all comes from the priced hypothesis tree.
Current State Process Map From Three Sources
Fill in the form and your workspace opens with the work already underway.