Operon.Cloud Workflow Economics
Case-level cost control · modeling · target execution
Finance and operations executives
Prior authorization · 1,000 of 2.5M Cases a year
Platform Orientation
What if the granularity of operational control, one Case at a time, met the rigor of financial planning and cost control, from the model to the monthly variance?
Finance reports the average. Operations runs the queue. Neither can say what one Case cost, or why.
an allocation, a quarter late
Cost per Case went from $40 to $38. Was it mix, volume, rework or the new vendor? On 2.5 million Cases that is $5M nobody can attribute; the ledger holds totals by cost center, not causes by workflow step.
The team sees queues, re-entries and hand-offs every day and has no standard cost for any of them. The manager running Clinical Review has no figure for Clinical Review.
An AI feature or a vendor rate promises savings per Case. The promise never meets the recorded occurrences it depends on, so it is never confirmed or denied.
Both rooms are looking at the same workflow. What is missing is neither data nor discipline. It is a shared price on each recorded fact, kept apart from the target it is judged against, so that finance’s variance and operations’ rework become the same number.
Finance closes with one number, a quarter later. Operations sees every Case as it moves, today.
of about 2.5M a year
Every visit, hand-off and action is already written down as the work happens, with who did it and when. Workflow Economics adds one thing to that record, a price on each kind of fact, and the record does the rest.
The record holds four kinds of fact. Give each kind one price, and the cost of every Case falls out of its own history.
$31.50 per Case on the sample
Nothing new is captured, nobody fills in a timesheet: twelve prices meet the facts operations already records, Case by Case. Finance would call it activity-based costing without the time study; operations, the first cost figure that names a cause.
Finance owns the price list. Operations owns the workflow. The record joins them.
sourced and signed, not guessed
Replay last year through the same price list before you trust this year.
Every price is a rate times a duration, so it can be read in dollars or in minutes. The CFO reads one, the CMO the other, from the same rows.
A price list that finance signs and operations recognizes is the contract between the two rooms. Everything downstream, every gap and every business case, inherits its credibility.
Control the cost of today’s work. Model the change. Then run the work against the target you set.
one record, any assumptions
This is the family of problems Workflow Economics is built for: case-level cost control, planning and modeling of process change including new systems and AI, and target execution models with financial controls tracked against benchmarks. The next pages show the building blocks, one per conversation the two rooms need to have.
A price list can be as coarse or as fine as what you know. Start flat, refine by workflow step, then by action.
operations fills the rows
Everything no step captures: overhead, postage, external fees. Enter 0 once the steps below carry it all.
A standard cost each time a Case enters the step. Coming back pays it again.
A surcharge or a credit per occurrence, relative to the visit. Each action carries one of four review states; Unreviewed prices as zero and is flagged.
Review completeness, 6 of 8 actions here, tells both rooms how much of the workflow has actually been decided. It is honest about its own depth: a flat model is a legitimate first model, and every refinement is a decision someone signed.
Commit at the level you can own: one figure for the Case, a share for each workflow step, and a date.
operations owns the shares
The $15.00 for Clinical Review is a cohort average for that step, not a cap on any one Case. It gives the step's owner a number to run to, and gives the target a shape that the modeled cost can be laid against.
Lay the target and the modeled cost on one scale, and the gap explains itself workflow step by workflow step.
+$1.50 per Case
Finance reads the top row: 85% of the way from last year to the target, $1.50 short. Operations reads the middle row: Clinical Review is $1.40 over its share, and the reasons are named. Same scale, same numbers, two rooms.
The same ledger at every altitude: a month, a cohort of Cases, one Case.
the number never changes
Cost per Case by the month a Case started. Benchmarks are dotted lines; nothing is interpolated toward them. Total money by month is the same bars times Cases started, against Target × Cases as the implied budget.
Every count is a door: the 720 Cases with a Human Review, the 200 that came back. At 208,000 Cases a month those doors open on 150,000 and 42,000 Cases, sorted by cost.
"Why did August move?" is a finance question. "Why did this Case cost $38?" is an operations question. Both are answered from the same rows, so the monthly figure and the single Case can never disagree.
A 1,000-Case sample keeps the arithmetic legible. At 2.5 million Cases a year it is a budget line.
≈ 208,000 Cases a month
Budget vs actual, but by the month the work started and with every dollar traceable to a workflow step and an action. Work done later on those Cases is shown separately as effort in month, so a per-Case budget is never compared to a mixed one.
The $1.50 that looks like rounding on one Case is $3.75 million across the year, and the ledger says which step and which action it sits in. That is a budget conversation with a name attached, before the quarter closes.
The same $31.50, read as 19 minutes of whose time, and where those minutes are lost.
every price is a rate × minutes
Cost of effort, not cost of decisions. Whether the right Cases were approved or denied, and whether they met the 72-hour deadline, are quality and compliance questions. They live beside this ledger, not inside it. FTE figures assume 1,700 productive hours a year.
An efficiency mandate is stated in hours and headcount; a budget in dollars. Because every price is a rate times a duration, the same rows answer both, and the rework, exception and automation lines are the same three levers in either unit.
Exceptions and automation are priced per occurrence, so ROI becomes a count times a price, on record.
+$2.10 per Case · $5.25M a year
AI-assisted Review saved $2,000 on 400 Cases. Take it to 800 and the same price list says −$4,000, which is $10 million a year at 2.5 million Cases. That is the case for adoption, in units the next month's record will confirm or deny.
Adoption is a count. Savings are that count times a credit someone stated. Both sit next to each other, per action, so the automation conversation moves from a vendor's slide to your own ledger.
A cost model is only useful while both rooms trust it. These rules are how it earns that.
Modeled, not reconciled
Every figure carries the label. It explains the ledger; it does not replace it. Finance keeps the books; this tells them what is inside.
A visit happened or it did not. Only the price attached to it can change, and changing it reprices every Case, past and future, the same way.
An action nobody has reviewed contributes nothing and is flagged wherever it would have counted. Review completeness is shown, not implied.
Targets are never capped or redistributed for you. Nothing is interpolated toward a target. A month with no Cases reads unavailable, never $0.
When workflow steps or actions are added, removed or renamed, the price list says so and asks both rooms to decide. What still matches keeps its price; what is new arrives unpriced.
The model summary assumes one visit per step and each action once. Only the record produces results, and only for Cases that actually ran.
Each rule trades a little convenience for a number that can be defended across the table: nothing inferred, nothing smoothed, every assumption visible where its effect appears.
Now take an operation you already run.
and the first model is a day's work
Cost per Case, priced from the record · September 2026