← Academy

Health-check your programme before you submit it

A DCMA 14-point assessment read contractually: what each check means, the thresholds, and why a clean programme is a claims asset long before any dispute.

Schedule · 7 min read · updated 2026-06-12 · FIDIC 2017 Red Book references unless stated

1. Programme quality is a contractual matter

Under FIDIC 2017 Sub-Clause 8.3 the programme is a contractual deliverable with prescribed content, and it is the instrument through which an EOT is measured. A network riddled with missing logic, hard constraints and lags will be attacked the first time it is used to demonstrate delay — by then it is too late to fix the baseline.

The DCMA 14-point assessment is the de-facto industry screen for schedule quality. Running it before submission — and before each update is issued — turns the programme from a liability into the evidential backbone of every future time claim.

2. The structural checks — is the network real?

Logic (missing predecessors/successors, target ≤ 5% of incomplete activities): an activity without logic floats free of the network and breaks any critical-path argument. Leads (negative lags, target 0%): they distort float and are near-impossible to defend. Lags (target ≤ 5%): prefer a real activity over an invisible wait. Relationship types (target ≥ 90% finish-to-start): exotic links obscure the path. Hard constraints (target ≤ 5%): a constrained date that overrides logic severs cause from effect — the very thing a delay analysis must demonstrate.

In ControlsIQ
Schedule Health runs all 14 checks natively on your P6 .xer or MS Project export, with each threshold adjustable to your contract's or client's standard.

3. The float and duration checks — is the network honest?

High float (> 44 working days, target ≤ 5%): large float usually signals missing logic rather than genuine flexibility. Negative float (target 0%): the plan admits it cannot meet its own constraints — explain it or fix it before anyone else finds it. High duration (> 44 working days on incomplete activities, target ≤ 5%): long bars hide the real sequence; break them down where the work is near-term.

Invalid dates (forecast in the past, actuals in the future — target 0%) and missed activities (finished later than baselined, monitored against 5%) test whether the update discipline is sound; resources checks whether effort/cost is loaded where the contract requires it.

4. The path checks — does the critical path behave?

The critical path test pushes a critical activity and checks the completion date moves with it — if it does not, the network's spine is broken. CPLI (Critical Path Length Index, target ≥ 0.95) measures how realistic the remaining critical path is against the baseline finish. BEI (Baseline Execution Index, target ≥ 0.95) measures whether tasks are being completed at the planned rate. Together they tell you — and the other side — whether the programme is a credible forecast or an optimistic drawing.

In ControlsIQ
The Schedule Health report presents each check with its result, threshold and the failing activities listed — exportable for the submission record. Schedule Comparison then tracks drift between baseline and each update before it compounds.

5. Read the results contractually, then fix forward

A failed check is not an academic blemish — it is the cross-examination question you will face when the programme is used to prove delay. Fix structure before baseline acceptance (logic, constraints, long bars), watch the update-discipline checks at every data date, and record why any deliberate exception exists (a contract-mandated constraint, for instance). A programme that passes clean is worth more than most claim narratives.

Run this workflow — freeSee the platform

Educational content for construction professionals. This guide summarises common contract mechanics and industry practice; it is not legal advice, and contract forms differ — your contract’s wording, including its Particular Conditions, governs. ControlsIQ outputs are designed to support professional judgement, not replace it.