Coastline Bank: A Program That Will Not Land Its Benefits
Coastline Bank approved an 18-month, $22 million program to modernize its lending platform, structured as five linked projects. At the month-10 gate, earned value data shows CPI 0.86 and SPI 0.92, with an estimate at completion of $26.5 million. The build is delivered by a systems integrator under a fixed-price contract; advisory and integration support is billed on time and materials. Fourteen change requests have been approved and nine more are pending, with the vendor estimating the pending ones at $3.1 million. The original business case promised $9 million a year in operations savings from reduced manual underwriting and $4 million a year in avoided fraud losses, but no baseline was measured before the program started. At the same steering meeting, the retail bank director asked the program to absorb mobile loan onboarding so the bank can match a competitor's launch in nine months.
1. Interpret CPI 0.86, SPI 0.92 and an EAC of $26.5 million. What is the honest position, and what do you do about the nine pending change requests given the fixed-price contract?
Read the worked answer
Read the indices honestly. CPI 0.86 means 86 cents of planned work per dollar spent - roughly $3.1 million of the $22 million has bought less than intended. SPI 0.92 is behind but less behind than cost, which typically indicates rework or unplanned effort rather than pure schedule slippage. The $26.5 million EAC is the delivery team's own forecast; sanity-check it independently, because EAC produced under pressure tends to be optimistic. A bottom-up re-estimate of remaining work is worth two weeks of effort.
The fixed-price boundary. The build is fixed price, so scope growth should be a vendor change order priced against contract rates, not T&M drift. The advisory and integration line is where the $3.1 million is most likely hiding scope that belonged in the fixed-price baseline. Before approving any CR: map each one against the original statement of work; require the vendor to classify each as clarification, defect or new scope; and get an independent cost estimate on the three largest.
Position for the steering committee: do not approve the nine as a block. Approve the two or three that protect a regulatory or launch-critical date, defer the rest to a post-go-live release, and formally re-baseline against the validated EAC so future variance is measured against reality. Accepting a re-baseline is not failure; running an unacknowledged 20% overrun is.
2. The $13 million annual benefits case was never baselined. What do you do now, and who should own benefits reporting?
Read the worked answer
The honest position is that $13 million of claimed annual benefit is currently unverifiable, and continuing to quote it is the fastest way to destroy the program's credibility. Benefits without a baseline are aspirations.
Actions:
- Retro-baseline, and label it as retrospective. You can often reconstruct 12 months of history on the two claimed drivers - underwriting cycle time and manual touch rate, and fraud loss per $1,000 of lending - from operations data even though nothing was captured at kickoff. Say clearly in the document that it was reconstructed, by whom, and from which source.
- Publish a benefits register: one line per benefit, with an owner, a formula, a data source and a measurement date. A benefit with no owner is not a benefit.
- Split the $9 million. Distinguish hard savings (roles actually eliminated, contractable) from capacity release. Finance will only credit the first. If the business case assumed elimination and the plan is now redeployment, disclose that now rather than at post-implementation review.
- Re-baseline the fraud benefit against actual loss trends. External fraud conditions move independently of your platform, and you will be accused of claiming credit for them otherwise.
Ownership: benefits reporting should sit with the finance business partner, not the program manager. A PM cannot credibly certify their own benefits. If measurement lands at $6 million rather than $13 million, take that number to the steering committee yourself - discovering it in a post-implementation audit is materially worse.
3. The retail director wants mobile loan onboarding absorbed into the program within nine months. How do you respond?
Read the worked answer
Say no to the date, not to the idea.
Bring three facts to the table: the program is at CPI 0.86 with an EAC already $4.5 million over budget; nine change requests remain unresolved; and the lending platform itself is not yet stable. Adding a customer-facing mobile channel onto an unstable core is the standard way to convert a cost overrun into a public outage.
Process:
- Quantify the value at stake. Ask what missing the competitor's launch by nine months actually costs. Sometimes the answer is 'very little' once written down, and the request dissolves on its own.
- If the value is real, run it as a separate, independently governed project consuming the platform's APIs after go-live, with its own business case and its own funding line. Do not merge it into a program already in variance - it will both hide the existing overrun and be blamed for it.
- If it must ship in nine months, present three options with cost and risk: (a) separate project, later go-live; (b) descope an equivalent value of existing scope to hold budget and date - and make the steering committee choose what to cut, not you; (c) add funding and accept schedule risk on the core.
Do not absorb it silently. The common failure mode here is a program manager who accepts the scope to stay popular and then explains the resulting overrun to the audit committee eighteen months later. Your value in this meeting is the honest option set, not agreement.