What It Costs
Licence is the visible number and rarely the largest. The costs that appear in year two, and how to phase so value arrives first.
Analysis
Process mining business cases are built on licence cost and surprised by extraction.
The visible cost
Licensing, commonly by case volume, by process or by user. The model matters enormously at scale, and case-volume pricing on a high-volume process can dominate the bill.
Check how the count works: cases loaded, cases retained, or events. The difference is large.
Model against the projection, not the current volume.
The costs that surprise
Extraction, which is your own people and is the largest first-year cost. Access approval alone is frequently weeks.
Data quality remediation, which is a process project rather than a technical one.
Mapping maintenance, ongoing, as source systems change.
Ownership, permanently.
Consultation and assessment, where task mining is involved.
A second extraction request, which is why the first should ask for every field you might want.
The cost that is not in the budget
Analyst time on exploration, which is where the findings come from and is unbounded if nobody scopes the question.
Operational time validating findings, which is necessary and comes out of someone else's week.
Change implementation, which is the actual improvement work and is frequently assumed to be free.
Sizing the value honestly
From your own log, not from a vendor percentage.
Decompose cycle time and take the largest non-value component.
Estimate a conservative reduction, with the assumption stated.
Multiply by case volume, which the log gives exactly.
Separate the compliance findings, which are certain and specific, from the efficiency estimate, which is not. Mixing them makes both look like estimates.
What to commit to
Not a percentage.
A specific change in a specific measure, with a date and the same query afterwards.
"Median approval decision time from 6.1 days to under 2 within two quarters, same log and boundary."
Checkable, which is what makes the second phase fundable.
Phasing
Value should arrive before the full cost does.
The first process, narrow, delivered in weeks, with one finding acted on and re-measured.
Then the licence decision, informed by what the manual analysis could and could not do.
Then additional processes, each cheaper than the last if the pattern was recorded.
A programme that buys a platform and then spends a year extracting has no evidence for its second year, which is the most common way these end.
The comparison worth making
Cost of the first analysis done manually against the platform licence.
For one process the manual route frequently wins.
For five it does not.
Which is an argument for doing one manually first, and it also produces the requirement that makes the eventual purchase narrower and cheaper.
The case-volume pricing trap
The commercial detail that decides whether a high-volume process is affordable.
Ask exactly what is counted: cases loaded, cases retained, events, or users.
Ask whether historical reloads count again.
Ask what happens when you extend the window, which you will want to do.
Model against your actual volume and the projection, not against the pilot.
A per-case model on a million-case process can exceed the whole programme budget, and it is discovered after signature more often than before.
Phasing the spend
Value should arrive before the full cost does.
First process, manual, narrow, weeks. One finding acted on and re-measured.
Then the licence decision, informed by what the manual route could not do.
Then additional processes, each cheaper if the pattern was recorded.
Then continuous refresh and alerting, which is where the standing value is.
A programme that buys the platform and then spends a year extracting has no evidence for year two, which is the most common way these end.
The cost that is not in the licence
Three items that dominate the first year and appear in no proposal.
Extraction and mapping effort, which is your own people and is the largest single cost.
Data quality remediation, which is a process project rather than a technical one.
Ownership, permanently, or the log goes stale and the numbers become confidently wrong.
Plus, where task mining is involved, the consultation and assessment work, which is real effort and should be budgeted rather than treated as an afterthought.
Model these from your own first process before committing to an enterprise licence.