Demonstration Scenario This is a constructed illustration, not a client engagement. The project does not exist. Figures are internally consistent and modelled on patterns observed across real developments — they are not measurements of any actual project.
Case Study

Four findings, one index, and the difference nine months makes.

A 168-unit mixed-use rental, eleven months into a twenty-two-month build. The construction platform said the job was healthy. The deal said otherwise.

$84.5M
Total development cost
$58.0M
Construction loan · 8.4%
Month 11
Engagement begins · 58% complete
44 → 68 → 51
Risk index over 14 weeks
The Deal

What the project was underwritten to do.

Every score in this scenario is measured against this column — not against the contract, and not against the approved budget.

TermUnderwrittenWhy it matters
Programme168 units, mixed-use rental, secondary US metroScale sets the absorption assumption
Total development cost$84.5M · hard cost $61.2MThe denominator for every variance
Hard-cost contingency$3.06M (5.0% of hard cost)The buffer between a change and a capital call
Construction loan$58.0M at ~8.4% · interest reserve $4.6MCarry runs ≈ $406K per month fully drawn
Schedule22 months to TCOEvery week of slip is priced
Loan maturityMonth 30The hard date the deal turns on
Completion testTCO achieved and 60% leased (101 units) by month 30Two conditions, tracked in two different places, by nobody
Lease-up12 months to 95% from TCOThe back end that decides the return
What Was Connected

Five sources. One 30-minute call. Nobody filled anything out.

The project team was told nothing and asked for nothing. Every input below was already being produced.

SourceRead forHeld by
Construction management platformRFI and submittal aging, change events, daily logs, ball-in-court latencyThe builder
CPM schedule, native fileCritical path, float consumption, milestone varianceThe builder
Cost ledger & requisition packagesCommitted vs billed, contingency draws, anticipated cost reportThe owner
Loan documents & draw historyInterest-reserve burn, covenant thresholds, completion-test termsThe owner
Municipal filing portalSign-off status of record, agency queue positionNobody. It is public.

The last row is the one that matters. Four of these five sources are inside the project. The fifth is outside it — and it is the one that produced the finding no participant could have reported, because no participant was looking.

What The Engine Found

Four findings in the first six weeks.

Each one priced against the underwriting, named to a seat, and carried on the record from first signal.

Financial Intelligence · Week 2 Red

Contingency is being consumed 1.19× faster than work is being completed.

At 58% complete, $2.11M of the $3.06M hard-cost contingency had been drawn — 69% consumed against 58% of the work. The budget report showed contingency "within tolerance," which was true of the level and false of the trajectory. Projected forward at the observed rate, the buffer exhausts at 84% complete, around month 16 — leaving roughly six months of construction with no contingency at all.

Exposure$0.6M – $1.4M projected overrun, plus six months unbuffered
Responsible seatOwner's Rep · CM Preconstruction Lead
Schedule & Execution Intelligence · Week 3 Red

Billed percent-complete led independently verified percent-complete by nine points.

Requisition #11 billed 61% of hard cost cumulatively. Independent verification at the monthly walk put work genuinely in place at 52% — a $5.5M cumulative gap. Of the $4.2M requested in that cycle, $1.4M–$2.2M represented work not yet performed. The lender's inspector had signed the prior two draws without the discrepancy surfacing.

Exposure$1.4M – $2.2M in the current draw cycle
Responsible seatCM Project Executive · Lender's Inspector
Certificate Path · Week 6 — the finding no participant could report Red

Nine sign-off items shown as "on plan" had never been filed.

The certificate-of-occupancy path carried 41 sign-off items. The builder's closeout tracker showed 34 of them "on plan" and 7 open. ClaimZero checked all 41 against the municipal filing portal — the public record, not the tracker. Nine of the 34 had no filing of record. They had never been submitted.

Two of the nine carried agency review queues exceeding 90 days: the elevator final and the sprinkler hydrostatic sign-off. Recomputed against actual queue position rather than the tracker's assumption, the achievable TCO date moved 11 weeks later.

Exposure≈ $1.0M in carry alone (2.5 months × $406K) — before the completion-test interaction below
Responsible seatCM Closeout Manager · Expediter of record
Capital & Sales Intelligence · Week 6 Amber

Pre-leasing was running at roughly 60% of the underwritten pace.

Tour-to-application conversion and deposit velocity implied a lease-up of about 19 months to stabilisation against the 12 months underwritten. On its own this reads as a marketing problem to solve after opening. Against the loan it is not.

Exposure$0.4M – $1.2M if renegotiated in advance; materially more at maturity
Responsible seatSponsor · Director of Leasing

The two findings nobody was holding together.

  • Finding 3 pushed achievable TCO 11 weeks later than the closeout tracker showed.
  • Finding 4 put 60% leased roughly seven months after TCO rather than four.
  • Stacked, the completion test — TCO and 101 units leased — moved from month 30 to month 34. Four months past loan maturity.
Neither input was construction data. One lived in a municipal filing portal; the other in a leasing pipeline. No platform on the project held both.
The Index

The number moved because the evidence moved.

The Project Risk Index scores trajectory, not level. It rose as findings landed and fell as the owner closed them — which is the only behaviour that makes a weekly number worth reading.

Project Risk Index — weeks 1 to 14

0 = on plan · 100 = maximum exposure. Higher is worse.
8060 40200 44 68 51 PEAK · CERTIFICATE PATH FINDING WK 1WK 4WK 7 WK 10WK 14
View as table
WeekIndexWhat changed
144Baseline established from ingest
247Contingency trajectory flagged
357Billed vs verified gap confirmed at the walk
459Requisition held pending reconciliation
561Certificate-path audit opened
668Nine unfiled sign-offs confirmed against public record
767Nine items filed
864Two long-lead queues confirmed and scheduled
962Completion test re-forecast with lender
1063Absorption held below plan a further month
1158Extension option agreed at 25 bps
1255$900K soft-cost contingency reallocated
1353Draw cycle reconciled, no further gap
1451Trajectory negative for four consecutive weeks
What The Owner Did

Five decisions, each made before the deadline that would have forced it.

ClaimZero does not run the project. It produced the page; the owner made the calls.

Week 4

Held requisition #11 pending reconciliation. $1.6M deferred to the following cycle rather than funding work not in place. The CM was given the verification detail, not an accusation.

Week 7

Filed the nine missing sign-off items. Because this happened in month 11 rather than month 20, the 90-day agency queues were absorbed into remaining float instead of consuming it.

Week 9

Re-forecast the completion test with the lender — nineteen months before maturity. The conversation was a re-forecast, not a workout.

Week 11

Agreed a 90-day extension option at 25 basis points. Priced as an option on a performing loan, because that is what it was.

Week 12

Reallocated $900K of unspent soft-cost contingency to hard-cost buffer. Restoring roughly three months of runway to the exhausting line.

The Counterfactual

The same four findings, discovered in month 20.

Nothing about the project changes. Only the date of discovery changes.

What month 20 looks like.

  • The nine filings hit 90-day agency queues with no float left to absorb them. TCO lands past month 30.
  • The completion test is missed, not re-forecast. The lender conversation begins in default. Extension pricing is no longer 25 basis points.
  • Contingency exhausted four months earlier. Overruns are funded by an equity call, negotiated from weakness.
  • The billed-versus-verified gap has compounded across nine more draw cycles and is now a dispute rather than a reconciliation.
  • There is no contemporaneous record of what was known and when — so the argument about responsibility is reconstructed from memory.
The engine found nothing a competent team could not have found. It found it in month eleven.

What this demonstration does not prove

  • It is a constructed scenario. No such project exists. Figures are internally consistent and modelled on patterns observed across real developments, but they are not measurements of anything.
  • ClaimZero has no published error rate, and none is claimed here. Calibration accrues per engagement; until a meaningful sample exists, the honest position is that no statistical claim can be made.
  • A real engagement will surface different findings, in a different order, at different magnitudes. The four here were chosen because each demonstrates a different data domain, not because they are typical.
  • Two of the four findings depend on the owner granting access to loan documents and the underwriting. Where that access is withheld, those domains read as unavailable — never as clear.
Founding Owner Offer

Run this against a project you actually own.

We will read your active project files and show you what the same four domains return on a real development. Institutional sponsors can convert the demonstration into an onboarding pilot.