Execute · Projects in construction
Which projects are drifting, and which recovery works?
Execute turns the monthly cost ledger and progress report into a P80 cost and finish for every unit, then prices the ways out. A contractor's milestone date is treated as one piece of evidence, not the forecast.
Output: P80 cost and finish for each unit, updated from actuals; recovery options priced in money and previewed before commitment; change orders routed by your authority matrix.
Who it is for
For the people answerable for the finish date
Project controls teams, PMOs, project directors and lenders' engineers. Execute sits above Primavera and SAP; it reads what they export and adds the range they do not give you.
- Project controls
- PMO
- Project directors
- Lenders' engineers
- Owner's team
Questions it answers
- What will each unit cost and when will it finish, at P50 and at P80?
- Is remaining contingency enough for the P80, and what do pending claims do to it?
- Which contractor dates are optimistic, and are they moving between reports?
- Which unit drives the programme finish, and what does a day of delay cost?
- Would a second shift, crew moves or resequencing pay for itself?
What Capibud produces
A monthly review with the range attached
Earned-value measures are the starting point. Capibud blends them with a reference-class prior, simulates the risks that remain and checks the result against crew ceilings and calendar windows.
P80 cost and finish
For each unit and the programme, from correlated scenarios, with the chance of finishing within budget authority.
A verdict per unit
Proceed, watch, recover or escalate, with the reasons stated in plain terms beside each verdict.
Contractor date checks
Reported milestone dates against Capibud's P50 and P80, and how far the reported date moved across recent reports.
Contingency coverage
Remaining contingency against the P80 need, with pending change orders simulated at their likelihood of being granted.
Priced recovery
Second shifts, crew moves and resequencing priced in money and previewed against shared crew ceilings before anyone commits.
Routed change orders
Every change order follows the authority matrix, with maker-checker and cumulative-change rules.
Worked example
Phase 1 of an AI campus, sixteen months in
Illustrative demo — not client dataA synthetic hyperscale campus has six packages and a combined budget authority of $1,185M. At the September data date, CPI is 0.96 and SPI 0.95. Only 42% of 2,000 correlated scenarios finish within authority, and remaining contingency covers 73% of the P80 need.
The data-hall fit-out drives the finish: P80 about six months beyond target, at roughly $520,000 a day. Crews on that package are at 79% of what the work in hand needs, so the first recovery option to price is a second shift.
The same campus, from option choice to construction, is in the data-centre developer case.
| Package | Budget | P80 cost | P80 finishvs target | Verdict |
|---|---|---|---|---|
| Shell and core | $176M | $187M | Oct 2026Sep 2026 | Proceed |
| Long-lead electrical | $228M | $257M | Feb 2027Dec 2026 | Recover |
| 230 kV substation | $106M | $120M | Aug 2027Mar 2027 | Recover |
| Cooling plant | $150M | $177M | Mar 2027Jan 2027 | Escalate |
| Backup generation | $96M | $98M | Jan 2027Dec 2026 | Proceed |
| Data halls 1–3 fit-out | $325M | $392M | Feb 2028Aug 2027 | Escalate |
Illustrative demo — not client data. Synthetic execution history; one seeded engine run. Budgets are at award; package P80s do not sum to the programme P80.
Bars scale to 120 days. A reported date is evidence; Capibud's P80 also reflects productivity, remaining risks and crews.
How it works
From the monthly pack to a priced recovery
What it connects to
Recovery money comes from the same envelope
A recovery plan is a capital decision. In the committee it competes with new projects and capex lines, and it often wins: protecting a committed programme is cheap value.
Recovery as a candidate
Extra crews or shifts are pooled with every other use of capital.
The capital committeeDrift caught early
New actuals or a contractor's listed peer falling trigger a priced proposal.
The brainEvery change order
Who approved what, on which forecast, is kept in the hash-chained log.
The recordHonest limits
What Execute does not do
Execute is a forecasting and decision layer. It does not run the job site or replace your planning tools.
- It is not a scheduling tool. Critical-path networks stay in your planning software; Capibud reads their exports and ranges the result.
- Schedules arrive as exports. Capibud works with what P6, MS Project and your ERP export today (XER, XML, Excel); live connectors are rolling out.
- Garbage in, ranges out. If progress is over-reported, the forecast inherits it. Capibud flags inconsistencies; it cannot measure the site.
- Productivity windows are inputs. Monsoons, festival leave and turnaround draws on crews are modelled as you describe them, labelled as estimates.
Related
See it worked, by industry
A replay asks the hardest version of Execute's question: how early could the overrun have been seen? The data-centre example shows the everyday version, on monthly actuals.
How a megaproject replay works
Quarter by quarter on point-in-time public information, with no look-ahead, shown on one invented project.
Read the explainerA data-centre developer: keeping phase 1 on track
Six packages, contractor dates checked against P80 and a second shift priced.
Read the caseBring us the project that keeps slipping.
A Decision Sprint can take one recovery plan from your monthly reports to a priced, routed decision in six weeks.