Every recovery starts with the same admission: the baseline is no longer the truth. Refusing to re-baseline is not discipline — it is denial. The projects that slip furthest are the ones whose weekly reports insist everything is on track until the day the client's finance team notices the S-curve has flatlined for six weeks.
Earned Value Analysis (EVA) is not a Primavera feature; it is a management philosophy. Once you commit to measuring physical progress against planned progress in cost-weighted units, most of the theatre falls away.
Step 1: Reconcile the model
Bring the P6 schedule into agreement with what is actually on site. Update % complete against physical measurement, not perception. We send two engineers with tape measures and a printed WBS; anything they cannot see and count gets 0%.
This first pass typically finds 15–25% of reported progress to be aspirational. Rebar tonnage delivered but not fixed. Blockwork raised but not plastered. Cable pulled but not terminated. Each of these has been counted as 'progress' in weekly reports and each is invisible on site.
The reconciliation is uncomfortable. It usually produces a written apology from the project manager to the client. It also produces the only trustworthy schedule the project has had in months.
Step 2: Compute CPI and SPI honestly
Cost Performance Index below 0.9 and Schedule Performance Index below 0.85 signal that the delivery method itself needs to change — not just harder work. CPI × SPI is the leading indicator: values below 0.75 mean the project cannot recover through effort alone.
At that point the recovery plan needs one or more structural interventions: additional shifts, prefabrication of a critical package, replacement of an underperforming subcontractor, or scope descoping with the client's agreement. Pretending harder execution will close a 30% variance is malpractice.
Track EVA weekly, chart it, and share the chart with the whole delivery team. The number on the wall changes behaviour faster than any memo.
Step 3: Shorten the critical path, not the reporting
Add resources to the driving activities, sequence overlapping trades, and remove non-critical scope from the current phase. Optimism in a look-ahead is the enemy of recovery.
Fragmentation of the critical path is the single most useful recovery move. A 90-day façade activity split into three 30-day zones, each with its own subcontractor, is often faster than one continuous crew — provided the interfaces are designed and the hoisting plan supports it.
Overlap requires design maturity. You cannot overlap MEP first fix into blockwork if the coordinated services drawings do not yet exist. This is why recovery efforts frequently begin with an emergency BIM coordination push rather than more labour on site.
Step 4: Change the meeting cadence
Weekly progress meetings become daily 30-minute stand-ups focused on the next 72 hours. The look-ahead is three weeks, not eight. Actions have named owners and next-day due dates. Anything unresolved for 48 hours escalates to the project director.
This is exhausting for the team and non-negotiable during recovery. It typically continues for 8–12 weeks, at which point EVA metrics have stabilised and normal cadence resumes.
Step 5: Document the settlement narrative
Recovery projects almost always end in a variation and/or extension of time claim. The daily records collected during recovery — signed permit-to-work, delivery tickets, force majeure logs, RFI response times — are the evidence base for that claim.
Keep the file live from day one of recovery. Nothing invalidates a legitimate EoT claim faster than reconstructing it after the fact.
Reading the S-curve honestly
The planned S-curve is a promise made under conditions that no longer exist. Comparing today's earned curve against a stale baseline produces theatrical variances that demoralise the team and mislead the client. Rebaselining is not moving the goalposts; it is telling the truth.
We plot three curves on the same axes: original baseline (grey, dashed), current rebaseline (navy, solid), and actual earned (gold). The visual immediately reveals whether the current recovery plan is landing or drifting. Anyone in the project can read the chart in five seconds — that clarity is the point.
Weekly variance below 2% of planned earned value is normal noise. Above 5% for two consecutive weeks triggers a mandatory review of the recovery assumptions. Above 10% triggers a re-forecast and a formal notice to the client. These thresholds are written into our project execution plans, not left to judgement.
The subcontractor conversation
A recovery plan that depends on subcontractor performance without a subcontractor conversation is fiction. We convene individual meetings with every trade contractor on the critical path in week one, share the reconciled schedule, and negotiate specific commitments — daily output rates, crew sizes, working hours, and interface dependencies.
Commitments are written into supplementary agreements with financial consequences. A trade that agrees to 400 m² of blockwork per week and delivers 250 forfeits a defined amount per m² of shortfall, capped at 10% of the trade's contract value. Conversely, over-delivery earns a bonus at the same rate. The mechanism is symmetrical and it works.
When recovery isn't possible
Occasionally the honest answer is that the project cannot be recovered within the original commercial envelope. Recognising this early is the professional's duty; hiding it prolongs the pain and increases the eventual cost.
In those cases we present the client with a scoped set of options: extend the programme by X weeks at Y cost, descope package Z to hit the date, or terminate for convenience and re-tender the remaining scope. Each option carries a written risk assessment. The client's decision is documented and dated. Nobody enjoys these conversations. Every project that has had one has ended better than one that avoided it.