The migration that has used up every day of float
A 26-week data-centre migration is at the end of week 10. The approved critical path is: data mapping (5 weeks), interface build (8 weeks), migration rehearsal (3 weeks), user acceptance testing (4 weeks), cutover (2 weeks). The contract completion date is the end of week 24.
Data mapping finished at the end of week 7, two weeks late, because the legacy data owner was unavailable. Interface build therefore started at the beginning of week 8 and has completed three of its eight weeks of work; the build lead forecasts five more weeks, finishing at the end of week 15. All later tasks are still estimated at their baseline durations.
UAT needs four analysts for four weeks. Two of them are booked on another programme for all of weeks 19–22. Contractual delay damages are £15,000 per week, management reserve is £30,000, and a contract analyst costs £3,500 per week.
1. Calculate the forecast completion week and the float position at the end of week 10. What does this tell you about the project's schedule risk?
Read the worked answer
Forecast
| Task | Weeks |
|---|---|
| Data mapping (actual) | 3–7 |
| Interface build (forecast) | 8–15 |
| Migration rehearsal | 16–18 |
| UAT | 19–22 |
| Cutover | 23–24 |
Forecast completion is the end of week 24, against a baseline of week 22 and a contract date of week 24.
Float position
- Baseline float on this path = 24 − 22 = 2 weeks.
- Slippage absorbed so far = 2 weeks (build finishing week 15 rather than 13).
- Remaining total float = zero.
What it means
- The project is not yet late, but it has no protection. Any further delay of even one day on interface build, rehearsal or UAT becomes a breach and triggers liquidated damages at £15,000 per week.
- Zero float is not the same as healthy. The two-week slip was absorbed silently, which usually means the next slip will be too.
- The forecast is built on unchanged estimates for rehearsal and UAT, which is optimistic - late data mapping often means the data is dirtier than assumed, and UAT is the stage where that surfaces.
- A single-path add-up is not a substitute for re-running the network; if any later activities run in parallel, the critical path may already have moved.
Actions
- Re-run the schedule with the actual dates and confirm the critical path and float.
- Put a named owner and a daily checkpoint on interface build.
- Prepare, but do not yet execute, a recovery option (see next task).
2. Two of the four UAT analysts will be unavailable for weeks 19–22. Recommend how to handle this, showing your reasoning.
Read the worked answer
Option A - run UAT with two analysts
UAT is 4 analysts × 4 weeks = 16 analyst-weeks. With two analysts for weeks 19–22 you complete 8 analyst-weeks; the remaining 8 need the four analysts at 2 weeks, so UAT finishes end of week 24. Cutover then runs weeks 25–26: two weeks late, £30,000 in damages, plus two weeks of deferred business benefit and reputational damage with the client.
Option B - hire two contract analysts
2 × 4 weeks × £3,500 = £28,000, UAT completes on time, cutover at weeks 23–24. The funding would come from the £30,000 management reserve, which needs sponsor approval.
Option C - escalate to release the two analysts
The conflict is a programme-level resource allocation problem, not a project problem. The programme manager may be able to reschedule the other programme's work at little or no incremental cost.
Reasoning and recommendation
- Option A is superficially cheapest in cash terms but £30,000 of damages buys two weeks of slippage - and damages are not the whole cost. Deferred benefits, the credibility hit and the pressure to compress cutover all add to the real figure. Reject it.
- Option B at £28,000 is almost exactly the damages figure, so on a pure cost basis it is close to neutral. That is the wrong comparison: it buys certainty of the date, which is worth more than £28,000. But reserve is finite (£30,000 total, and other risks remain), and route B consumes nearly all of it.
- Option C is free if it works.
Recommendation: escalate for release of the two internal analysts first, with a firm deadline for an answer. If it is refused, authorise the two contract analysts against management reserve, with sponsor approval, and record the drawdown against the reserve. Do not let the decision drift past week 16 - the longer it waits, the more expensive every option becomes.
Also test the estimate: is four analysts for four weeks genuinely the resource profile, or could two experienced analysts plus two business users complete the same coverage? Levelling the profile is cheap; assuming it cannot change is not.
3. The sponsor asks for written assurance that the contract date will be met. What do you say?
Read the worked answer
Do not give a binary assurance - it would be false confidence. Give a forecast with a confidence statement and the conditions attached.
Draft position
- Current forecast: completion end of week 24, which meets the contract date.
- Confidence: low to moderate, not high. The forecast has zero float, depends on an unreleased resource decision, and rests on unrevised estimates for the two stages where late data usually shows up.
- Range: best case week 24 (on time, no further slip); realistic case week 25–26 if the analyst conflict is not resolved or rehearsal overruns; worst case week 28+ if rehearsal exposes significant data quality issues and UAT cycles.
Conditions for the assurance to hold
- Analyst availability resolved before week 16.
- Interface build delivers to the revised date of week 15 - this is now a tracked milestone, not a forecast.
- Rehearsal completes within three weeks; if it overruns, the recovery plan (fast-tracking rehearsal into UAT, with the rework risk quantified) is invoked immediately.
What I would also say to the sponsor
- The £30,000 management reserve is now the only buffer against schedule cost; ask for pre-approval to draw on it rather than coming back at the point of crisis.
- Delay damages of £15,000 per week understate the true cost of a late cutover; the decision on the analysts should be made on total cost, not on cash.
- The two-week slip on data mapping has never been formally reported as a schedule variance; it is being reported now because it has consumed all the float.
An assurance that says 'yes, week 24' with no conditions is a promise the project manager cannot keep and should not make.