Forecasting an ERP Rollout When Cost and Schedule Diverge
Horizon Retail is deploying an ERP system over a 10-month baseline with a budget at completion (BAC) of $2,400,000. At the end of month 5 the project manager reports planned value (PV) $1,200,000, earned value (EV) $1,020,000, and actual cost (AC) $1,175,000. Two critical-path configuration modules slipped when a key analyst left; to recover, the team has been running two premium-rate contractors at $310 per hour since month 4. The vendor that supplies the licensed test environment has reserved months 8–10 for Horizon and cannot move the slot without a six-figure fee. The sponsor wants a single forecast number and has suggested compressing system testing from ten weeks to seven to get back on track. Any scope change must pass the weekly change board, which has already rejected one request to descope the fixed-asset module on audit grounds. Contractors are billing 55-hour weeks and the internal team is showing fatigue-related defects in unit testing.
1. Calculate CPI, SPI, and a CPI-based EAC. What do these figures tell you, and what does the TCPI imply about the sponsor's $2.4M target?
Read the worked answer
Metrics
- CPI = EV / AC = 1,020,000 / 1,175,000 = 0.87
- SPI = EV / PV = 1,020,000 / 1,200,000 = 0.85
- EAC (CPI-based) = BAC / CPI = 2,400,000 / 0.87 ≈ $2,759,000
- VAC = BAC − EAC ≈ −$359,000
- TCPI to BAC = (BAC − EV) / (BAC − AC) = 1,380,000 / 1,225,000 ≈ 1.13
Interpretation. Both indices are below 1: the project is behind schedule and over budget. Cost performance is slightly worse than schedule performance and is deteriorating, because premium contractors inflate AC without a proportional EV gain yet. The SPI is a cost-weighted indicator, so it understates or overstates schedule position depending on where the money sits; the critical-path slip is the more reliable schedule signal.
The TCPI of 1.13 says the remaining work must be delivered 13% more efficiently than the original plan to land on the $2.4M BAC. Given premium labour rates and already-visible fatigue defects, that is not credible. The honest message is that BAC is unachievable without descoping, and the CPI-based EAC is itself optimistic if testing is compressed and rework follows. Present a range: ~$2.76M as the likely case, plus a bottom-up sanity check from the remaining work packages.
2. The sponsor proposes cutting system testing from ten weeks to seven. Evaluate this as a recovery action.
Read the worked answer
Do not treat testing compression as the primary recovery lever.
- It does not fix the cause. The slip originated in configuration. Compressing a downstream phase buys calendar time but does not create the missing configuration output; testing cannot run on modules that are not configured.
- The binding constraint is the vendor slot, not test duration. Months 8–10 are reserved and immovable. Cramming ten weeks of execution into seven inside that window raises defect-escape risk on a system with audit implications.
- It is fast-tracking. Overlapping configuration completion with test execution only shortens the schedule if the overlapped work does not have to be redone. Configuration defects will surface late and cost more to fix.
- It is a baseline change. Reducing planned test coverage changes the quality baseline and belongs in front of the change board, with the risk documented and accepted by the sponsor in writing - especially since the board has already protected scope on audit grounds.
Better alternatives: finish configuration behind a quality gate, then use the fixed test slot at full capacity from day one by pre-staging test data and scripts; parallelise independent test streams rather than cutting coverage. If the sponsor still insists, quantify the defect-escape exposure and record the acceptance with a named risk owner.
3. Recommend a corrective action plan for the next change board, including what to do about the contractors.
Read the worked answer
Recommended plan, in order:
- Re-forecast transparently. Update EAC to ~$2.76M and issue a revised schedule showing the realistic finish. Do not let the organisation keep planning against $2.4M.
- Contain contractor spend. Keep the two contractors only on critical-path configuration with a weekly deliverable gate; return non-critical configuration to internal staff. Two contractors at 40 hours each and $310/hour cost roughly $62k per week - a run rate the CPI cannot absorb for long.
- Recover time on the critical path only. Parallelise work that genuinely can be parallelised without rework, and pre-stage test data so the fixed slot is used at full capacity.
- Protect quality. Insert a configuration peer review before each module hands off to testing; fatigue defects are already appearing in unit tests.
- Force the trade-off decision. Give the change board two options: (a) approve a revised budget of ~$2.76M and keep scope, or (b) descope a defined module to protect the budget. The sponsor owns that choice, not the PM.
- Update the risk register and reserves with the vendor-slot constraint, the contractor dependency, and the schedule risk, and reflect the revised EAC in the cost baseline once approved.