The Cost-Benefit Analysis projects $2,530,000 of annual benefit against a $11,582,050 investment, ramping 50% / 80% / 100% across the first three years. This plan makes that curve falsifiable: what each benefit is as a number, which operating department it lands in, who is accountable for declaring it, and when it first becomes measurable. Status as of 16 Oct 2026.
1. The structural problem this plan exists to solve
The program spends $11,582,050 from IT and program capital. Of the $2,530,000 in annual benefit, only the $380,000 legacy decommissioning line returns to a budget the program controls. The other $2,150,000 — claims labour efficiency, rework reduction, compliance exposure, retention — lands in Claims Operations, Compliance and Customer Service. None of those departments funded the program, and none of them was accountable for its business case.
That asymmetry, not measurement difficulty, is the usual reason internal benefit cases go unverified. The department that spent the money cannot see the benefit in its own numbers. The departments that can see it were never asked to commit to it. Everyone is behaving reasonably and the benefit still goes unclaimed.
2. The five benefits
Values are the CBA's, unchanged. What this document adds is the department the benefit lands in, the accountable executive there, and the first date the number can honestly be read.
| ID | Benefit | Annual | Measure | Lands in | Accountable owner | First measurable | Confidence |
|---|---|---|---|---|---|---|---|
| BEN-01 | Labor Efficiency (Claims Processing) Reduced manual processing time from automation |
$1,200,000 | Claims processed per FTE-hour, and total claims-processing FTE budgeted in Claims Operations | Claims Operations | J. Boudreaux (VP, Claims Administration) Prepared by F. Jones |
Q4 2028 | Medium |
| BEN-02 | Error / Rework Reduction Fewer claims errors and appeals |
$450,000 | First-pass claim adjudication rate, and appeals volume per 10,000 claims | Claims Operations | J. Boudreaux (VP, Claims Administration) Prepared by R. Whitfield |
Q3 2028 | Medium |
| BEN-03 | Compliance Risk Avoidance Risk-adjusted avoidance of regulatory penalty exposure |
$300,000 | Open regulatory findings attributable to platform limitations, and SOX control exceptions | Compliance / Internal Audit | N. Sharma (Compliance) Prepared by G. Fenwick |
Q2 2028 | Low |
| BEN-04 | Legacy System Decommissioning Savings Retired legacy license/maintenance fees |
$380,000 | Legacy platform licence and maintenance invoices confirmed terminated | IT Operations | W. Donnelly (Vendor / Procurement Manager) Prepared by M. Alvarez |
Q1 2028 | High |
| BEN-05 | Customer Retention / Satisfaction Faster claims turnaround reduces churn (conservative) |
$200,000 | Member churn rate and claims-turnaround NPS, against the same quarter of the prior year | Customer Service / Marketing | V. Alaoui (Customer Experience Manager) Prepared by L. Bergström |
Q4 2028 | Low |
| TOTAL | Five benefit categories | $2,530,000 | Matches the Cost-Benefit Analysis exactly. Net of $950,000 annual ongoing cost, full run-rate is $1,580,000. | ||||
Confidence describes how well a benefit can be evidenced, not how likely it is to occur. Compliance risk avoidance is Low confidence and entirely plausible — an avoided penalty leaves no entry anywhere.
3. Holding the realization curve to account
The CBA does not assume benefits arrive at full value on day one. It assumes a ramp, and the whole payback case depends on it:
| Period | Realization | Gross benefit | Ongoing cost | Net |
|---|---|---|---|---|
| Year 1 | 50% | $1,265,000 | $950,000 | $315,000 |
| Year 2 | 80% | $2,024,000 | $950,000 | $1,074,000 |
| Year 3 | 100% | $2,530,000 | $950,000 | $1,580,000 |
Stated as a curve it looks like prudence. Stated as a commitment it is testable: at the end of Year 1 the program must be able to show it is running at half of $2,530,000, and if it cannot, the ten-year case has already moved and the organisation should know that in 2028 rather than in 2032. Each annual review therefore reports actual realization against the curve as a percentage, and a miss is reported as a miss — not absorbed by moving the curve.
4. Why the enrollment cycle governs every measurement date
Assumption A-004 fixes go-live before open enrollment, because the date cannot move. The same cycle that makes the date immovable also makes naive measurement worthless: claims and enrollment volumes move several-fold between open enrollment and a quiet month. A claims-per-FTE-hour figure taken in November against a baseline taken in June measures the calendar, not the platform.
Two rules follow, and they are the reason the “first measurable” dates in section 2 sit where they do:
| Rule | Consequence |
|---|---|
| Every volume-sensitive measure compares to the same period of the prior year, never to the immediately preceding period. | Requires a full pre-cutover year of baseline data, which exists in the legacy platform and must be extracted before decommissioning. |
| The first open enrollment after go-live is excluded from benefit measurement. | It runs on a platform that is ten weeks old, with staff new to it. It measures the cutover, not the capability. The first clean full-cycle comparison is the second open enrollment — Q4 2028. |
This is also why D-002 matters beyond testing: pre-funding a third conversion dry run at $150,000 before it was known to be needed protects the go-live date, and the go-live date is what puts the first clean measurement window inside a known enrollment cycle rather than straddling two.
5. The program closes before its own benefits exist
Go-live is 24 Aug 2027. Program closeout is 2 Nov 2027 — ten weeks later. At closeout, zero of the five benefits has a defensible annual reading. This is normal and is not a planning failure; it is simply what happens when a program's job is to deliver a capability and the capability then takes years to pay back.
It does mean two things must be true at closeout, and they are the actual deliverables of this plan:
| # | What must be true at closeout | Owner |
|---|---|---|
| 1 | The full pre-cutover baseline year has been extracted from the legacy platform and stored somewhere that survives decommissioning, with its extraction method documented. | T. McCormick |
| 2 | Each receiving department has accepted its benefit in writing, including the BEN-01 declaration of whether the capacity saving becomes a headcount reduction or absorbs growth. | C. Tyrrell → S. Ryan (PMO) |
| 3 | The measurement definitions and queries run inside the operating reporting stack, not in program tooling that is decommissioned with the program. | J. Albert |
| 4 | An owner is named for the Year 1, 2 and 3 benefit reviews, with the reviews scheduled in the governance calendar before the program's own governance dissolves. | S. Ryan (PMO) |
The Closeout Report records these four as a handover status. It does not report a benefit figure, because at 2 Nov 2027 there is not one to report.
6. What is actually difficult about each line
Benefit registers fail by treating every line as equally measurable. These are not.
Labor Efficiency (Claims Processing)
$1,200,000/yrAt $1,200,000 this is 47% of the entire benefit case and the one most likely to be claimed without being realized. The saving is roughly 12–15 FTE of claims-processing capacity. It converts to cash only if Claims Operations actually reduces budgeted headcount or absorbs volume growth without adding to it. If neither happens the program has produced spare capacity, which is a real operational good and not $1,200,000. This plan asks Claims Operations to declare, before go-live, which of the two it intends — because that decision cannot be made retrospectively without it looking like a justification.
Error / Rework Reduction
$450,000/yrMeasurable from data the platform already produces, which makes it the most tractable line. The trap is the conversion boundary: I-004 established that where the specification and the legacy code disagree the legacy behaviour is the de facto requirement. Some existing “errors” are therefore encoded business rules, and correcting them will move the first-pass rate for reasons that have nothing to do with quality.
Compliance Risk Avoidance
$300,000/yrA risk-adjusted avoidance. Nothing appears in any ledger when it works, so the honest evidence is the finding and exception record, not a dollar figure. Reported as a count with the risk-adjusted value shown as derived rather than observed — the CBA is explicit that this is exposure × probability, and the plan does not launder that into a saving.
Legacy System Decommissioning Savings
$380,000/yrThe only line here that is close to certain, and the only one that produces cash in the same budget that spent it. It depends on one thing: the legacy platform being switched off rather than left running in parallel “just in case”. Parallel-run is the normal outcome when a conversion is not fully trusted, so the decommissioning date is tracked as a milestone in its own right, not assumed to follow go-live.
Customer Retention / Satisfaction
$200,000/yrThe CBA already calls this conservative and it should stay that way. Churn moves with premium pricing, network changes and competitor behaviour far more than with claims turnaround. It is kept in the case because excluding it would understate a real effect, and held to a low confidence rating because attributing any part of it to this program is genuinely hard.
7. Open gaps
| ID | Gap | Consequence | Owner |
|---|---|---|---|
| BG-01 | No receiving department has yet accepted its benefit in writing. The CBA was approved by the Steering Committee as a program business case; the operating departments named in section 2 were consulted but hold no commitment. | The largest line, BEN-01 at $1,200,000, has no owner obliged to report it. Without acceptance before go-live, benefit reporting after closeout is voluntary. | S. Ryan (PMO) |
| BG-02 | The pre-cutover baseline year has not been extracted, and the legacy platform is decommissioned as part of BEN-04. | Extracting the baseline and decommissioning the source are the same event in opposite directions. If sequencing is not enforced, four of five benefits lose their comparator permanently. | T. McCormick |
| BG-03 | No benefit review is funded after 2 Nov 2027. The program's budget ends with the program. | Year 1, 2 and 3 reviews have no cost centre. A review nobody is funded to run is a review that does not happen. | S. Whitcombe (VP, Finance) |
| BG-04 | Attribution. Six integrations go live together, and other initiatives touch the same operational metrics in the same period. | Benefit claims cannot be cleanly separated from concurrent change. Every report states what else changed in the period; this reduces but does not remove the problem. | E. Kowalski |
Risk R-001 already tracks the program's dominant delivery risk. The gaps above are deliberately not folded into it — they are benefit-realization risks, they materialise after the program's risk register has been closed, and rolling them into a register that stops being maintained at 2 Nov 2027 is precisely how they get lost.