- The question it answers
- Which records an extension of time claim actually needs, what each one proves, and precisely which analysis methods its absence forecloses.
- The compound rule
- The method available to you is a function of the records you kept. Record failure is not a presentational problem, it removes the strongest methods and pushes the claim toward the ones tribunals trust least.
- The six categories
- SCL Delay and Disruption Protocol, 2nd Edition (February 2017), Guidance Part B s.1 and Appendix B: programme records, progress records, resource records, costs records, correspondence and administration records, contract and tender documents.
- The single biggest evidential loss
- Losing the contemporaneous update series. It removes every dynamic and windows method at once and forces recreated updates (AACE MIP 3.5) or as-built-only analysis: both expensive, both inviting the "reconstructed for litigation" attack.
- What unexplained logic changes cost
- Alares Construction, Inc. v Department of Veterans Affairs, CBCA 6149, 7071, 7597 (21 March 2025): schedule logic unsupported or contradicted by contemporaneous project documentation, unexplained logic changes in the monthly updates, adverse inferences drawn. 765 days claimed. 218 awarded.
- Progress is not completion
- George Sollitt Construction Co. v United States, 64 Fed. Cl. 229 (2005): delay affecting only non-critical activities does not delay completion. A record that cannot distinguish the two cannot support an extension of time at all.
- Schedule validation, and its limits
- The DCMA 14-Point Assessment: section 4 of DCMA-PAM-200-1, not a standalone document: is a quality screen, not a legal standard, say so when you use it. And the recorded negative finding: no reported judgment decides a case on "float suppression" as a named doctrine. It is an evidential and credibility attack on the baseline and the updates, not a cause of action.
- Retention horizon
- URS Corporation Ltd v BDW Trading Ltd [2025] UKSC 21: Building Safety Act 2022 s.135, 30 years retrospective and 15 years prospective, for Defective Premises Act 1972 claims and claims dependent on them. Not a universal retention rule: it reaches delay records on projects with that exposure, and the horizon is set from the contract, the governing law, the limitation regime and the defects exposure, not from a single number.
Records get written about as hygiene: keep good ones, be organised, you will thank yourself later. That undersells the problem by an order of magnitude. The records a project keeps decide which delay analysis methods are available to it in a dispute, and that set is fixed long before anyone knows there is a dispute.
By the time a claim is being prepared, the choice of method has already been made. It was made by the project team, months or years earlier, without knowing they were making it: in the decision to issue programmes as PDFs, in the monthly update that was never statused, in the progress field that captured percentage complete instead of actual start and finish dates. The forensic decision was taken by people who were not thinking about a claim at all.
Which means the useful question on a live project is not "are our records good?" It is: given what we are actually holding today, which methods can still be run, and what would we have to start doing this month to add one back? That question has a determinate answer, and it maps directly onto the four-element entitlement test every extension of time claim has to satisfy: element two, causation to completion, and element four, extent.
Which delay analysis methods have your records already ruled out?
Work from the record set inward, not from the facts outward. A baseline and an as-built alone support only the as-planned versus as-built family. A reliable statused update series supports windows analysis, which is the best-evidenced position available. No accepted baseline removes the modelled additive methods in practice. The table below is the payload of this subject, and every row is a decision a project can act on inside a month.
| Record level actually held | Methods it makes available | What its absence forecloses | The decision available this month |
|---|---|---|---|
| Native baseline programme, formally accepted: logic, constraints, calendars, resource loading, narrative | The datum for every method, and the float profile | Impacted as-planned (SCL method 1 / AACE MIP 3.6) and time impact analysis (SCL 2 / MIP 3.7) become unavailable in practice. The float argument collapses for both sides (SCL Protocol 2nd Edition (February 2017), Core Principle 9) | Get acceptance in writing, or record in writing that acceptance was refused, and why. The employer's standing defence is that the baseline was never accepted, so the contractor's own plan proves nothing |
| Contemporaneous statused updates: dated, with data date, progress, logic changes | SCL time slice windows (AACE MIP 3.3 / 3.4), the best-evidenced position there is | Recreated updates (MIP 3.5) or as-built-only methods. This is the single biggest evidential loss available on a construction project | Status the update monthly and issue it. An update that exists but was never issued to the other side is worth a fraction of one that was |
| Native schedule files, not PDFs | Every CPM-based method | A PDF of a programme cannot be interrogated for logic, constraints or float, so it supports none of them | Change what the monthly submission contains. This is a five-minute administrative decision with a forensic consequence measured in years |
| A narrative with every update, explaining each logic change | Windows analysis that survives cross-examination | Unexplained logic changes are the single most damaging artefact in a record set, and they draw adverse inferences (Alares, below) | Add a logic-change log to the monthly report template: what changed, why, and on whose instruction |
| As-built dates recorded as actual start and finish | As-planned versus as-built windows (SCL 4 / MIP 3.2, with the gross non-windowed comparison at MIP 3.1). Collapsed as-built (SCL 6 / MIP 3.8, 3.9) | Progress captured as percentage complete gives no defensible as-built, and the analysis built on it is discounted (George Sollitt, below) | Change the field the site team fills in. Percentage complete is a reporting convenience, not a record |
| Superseded baselines and every accepted revision, preserved | Multiple-base modelled methods (MIP 3.7 / 3.9) | Only one base survives, and later re-sequencing reads as concealment rather than management | Lock and archive each revision together with the correspondence accepting it |
| Daily reports, site diaries and manpower returns | The fact layer under as-planned versus as-built windows. The primary check on the schedule updates | Windows analysis becomes assertion, the pacing defence fails, a disruption claim collapses into a global claim | Record manpower by trade by area, daily. "Good progress made" is not a record |
| RFI and submittal logs dated on both sides | Response-time analysis, which is how design-caused delay is actually proved | "We were waiting on information" has no critical-path anchor | Add the response-date column, then report average response time monthly. Response time is the metric |
| No accepted baseline and no update series | Nothing reliably | Total time claim territory. The weakest form of extension of time claim there is | The only fix is to start now. A rebaseline formally issued this month is a datum for every window after it, even if the months before it are lost |
The dependency runs one way. You cannot decide in the dispute to have kept updates during the works. Which method survives cross-examination is a question about the past, and the choice among the six SCL methods and the nine AACE method implementation protocols is real only within the set the record permits.
SCL Delay and Disruption Protocol, 2nd Edition (February 2017), Core Principle 1 recommends the parties agree the record regime in writing before or at the outset of the works: record types, who produces and checks them, frequency, format, ownership and storage. Almost nobody does this. It is the cheapest risk mitigation available on any project, and it belongs in the contract, not in the project execution plan.
Why the schedule update series is the load-bearing record
The contemporaneous schedule update series carries the analysis. Every other record corroborates it. Dated updates with a data date, statused progress and logic changes are the only record that shows the critical path as it actually shifted, which is what a dynamic or windows method exists to demonstrate. Their absence forces recreated updates or an as-built-only analysis. Their presence, where logic changes without explanation, is worse than useless, it hands the other side an adverse inference.
Alares Construction, Inc. v Department of Veterans Affairs, CBCA 6149, 7071, 7597 (21 March 2025) is the current textbook illustration. The contract required monthly updates reflecting activity and event changes and accurate actual-versus-planned progress. The Board found the contractor's schedule logic unsupported or contradicted by contemporaneous project documentation, identified unexplained logic changes in the monthly updates, adopted the government expert's critical-path analysis and drew adverse inferences. The failure to maintain the updates cost the contractor the benefit of the doubt on a claim measured in hundreds of days.
One caution on citing Alares, because it is frequently cited for the wrong proposition. It is sometimes offered as a change-order release case. The Board upheld the reservation of rights in modification P00005. The claim then failed on concurrency, which is an independent ground. A good reservation of rights preserves a claim. It does not prove one, and it does not repair a record set.
Two further points about schedule records that practitioners routinely get wrong.
Progress is not completion, and a record that cannot tell them apart supports nothing. George Sollitt Construction Co. v United States, 64 Fed. Cl. 229 (2005) held that a delay affecting only non-critical activities does not delay completion, and separately discounted the expert's analysis for missing baseline schedules, absent monthly updates during startup and completion, progress reported as percentages rather than actual start and finish dates, and inaccurate critical-path activity dates. Those two holdings belong together. Only delay to completion generates an extension of time (SCL Protocol 2nd Edition (February 2017), Core Principle 6 and Definitions). Non-critical delay generates, at most, a disruption or loss-and-expense claim. A record that logs "activity late" without actual dates and without a critical path cannot make that distinction, so it cannot support an extension of time at all, however many pages of it there are.
Nobody can invent an update after the fact. In Premier Modular Ltd v Maidstone and Tunbridge Wells NHS Trust [2026] EWHC 1404 (TCC), an NEC4 adjudication decision was refused enforcement in part because the adjudicator relied on an "updated" Accepted Programme when both parties agreed the programme had not been updated. Under NEC4 ECC a compensation event assessment stands or falls on the Accepted Programme. NEC4 ECC clause 63.5 compares planned Completion before the event with planned Completion after it, and there is nothing to compare if the programme was never accepted or never updated. Note the numbering while you are here: NEC4 programme retention is clause 50.5, not clause 50.3: 50.3 was NEC3, and commentary still using it is out of date.
What schedule validation actually checks, and what it does not
Schedule validation is a quality screen applied to the structure of a CPM network: logic, lags, constraints, float distribution, duration outliers, open ends. It tests whether a programme is built in a way that produces meaningful float and critical path outputs. It does not test whether the programme is a fair representation of how the works were planned or performed, and it decides nothing about entitlement.
The de facto first move is the DCMA 14-Point Assessment, which is not a standalone publication but section 4 of DCMA-PAM-200-1, the US Defense Contract Management Agency's Earned Value Management System Program Analysis Pamphlet: logic, leads, lags, relationship types, hard constraints, high float, negative float, high duration, invalid dates, resources, missed tasks, the critical path test, CPLI and BEI. It is genuinely useful and it is also a quality screen, not a legal standard, treating it as the latter is a mistake in both directions. A programme that fails several checks is not thereby non-compliant with the contract or inadmissible. Plenty of honest programmes on complex projects fail the high-duration and high-float tests for reasons that have nothing to do with gaming. A programme that passes all fourteen is not thereby validated as a truthful account of the works. The checks are structural, and a structurally clean network can still encode a sequence nobody ever intended to build.
What the screen is actually looking for, on either side, is the family of gaming artefacts:
- Artificial constraints: "Finish On", "Mandatory Start", "Start No Earlier Than", that mask or destroy float and manufacture criticality
- Excessive or negative lags used as pseudo-durations
- Preferential (soft) logic presented as physical logic, creating a critical path that does not reflect physical necessity
- Duration padding, time risk allowance buried inside activity durations rather than declared. Under NEC4 ECC clause 31.2 the programme must show float and time risk allowances separately, so on an NEC4 contract this is a direct clause 31.2 breach, not merely poor practice
- Open ends and dangling activities, which corrupt the total-float calculation
- Resource-levelling artefacts creating hidden dependencies
- Multiple critical paths and near-critical compression, so that any event at all can be portrayed as critical
- Baseline reissues that quietly re-sequence after the event, without a narrative
State the negative finding plainly when you run this argument: there is no reported judgment deciding a case on "float suppression" as a named doctrine. It is not a cause of action and it is not a defence. It operates as an evidential and credibility attack on the baseline and the updates, available to either party, and it succeeds or fails on whether the tribunal ends up believing the programme. Pleading it as though it were a recognised legal category invites the correction from the bench.
The float point underneath all of it is SCL Delay and Disruption Protocol, 2nd Edition (February 2017), Core Principle 9: identification of float depends on a properly prepared and regularly updated programme. Where that does not exist, the float argument collapses for both sides. The contractor cannot show the event consumed the last of the float, and the employer cannot show float remained.
Daily reports, manpower and when the impact actually began
Daily reports and manpower returns fix the moment an impact started, and they do it in a way no later narrative can. Labour by trade and by area, plant on site, weather, areas worked, obstructions encountered, instructions received: recorded the same day, by someone with no view about a claim. That is the fact layer underneath every as-planned versus as-built window, and it is the primary check on whether the schedule updates are telling the truth.
In practice (and this is a working observation from the record patterns in this corpus rather than a proposition from authority) the drop in manpower is often the first hard signal of when an impact actually began, and it usually predates the correspondence by weeks. A trade that walks off a floor because the preceding work is not ready shows up in the labour return before anyone writes a letter about it. The letter is advocacy written with a purpose. The labour return is a bookkeeping artefact created for payroll. Tribunals know the difference, and they weight it accordingly.
The same contemporaneity point decides pacing. A slowdown recorded at the time (in the schedule narrative, the resource plan and the correspondence) is a pacing election. The identical slowdown asserted afterwards is contractor delay, and it becomes the other side's case on concurrency, where the contemporaneous record beats the method label every time. Pacing has to be evidenced as an election made at the time. There is no retrospective route to it.
Four other categories in the same fact layer, each with a specific proof job:
- RFI and submittal logs, dated on both sides. Issue date, response date, response time, criticality. In Walsh Construction v Toronto Transit Commission, 2024 ONSC 2782, the owner's own consultant's finding that average RFI response time "was also excessively high" materially supported the contractor. Without the log, design-delay causation has no critical-path anchor.
- Weather records: site-specific plus a long-run meteorological baseline. "Exceptionally adverse" and "unusually severe" are comparative tests: FIDIC Red Book 2017 (reprinted 2022), Sub-Clause 8.5(c) exceptionally adverse climatic conditions. AIA A201–2017 §15.1.6.2. Log the baseline at tender, not at claim. Without it the claim reduces to ordinary contractor-risk weather.
- Procurement and long-lead logs: order, approval, fabrication, delivery, inspection dates. Frequently the decisive category on equipment-heavy EPC work, because it is what separates contractor-side procurement failure from employer-side approval delay. Absent it, the employer attributes the whole overrun to contractor procurement and the contractor has no rebuttal.
- The correspondence, notices and instructions register. This proves element three of the entitlement test and the timing of the employer risk events. It matters most where notice operates as a condition precedent, and it also destroys the "the employer knew anyway" fallback, notice must emanate from the claimant (Steria Ltd v Sigma Wireless Communications Ltd [2007] EWHC 3454 (TCC)).
One further category is becoming an expectation rather than a bonus: timestamped photographs, drone capture, 4D and reality capture, as independent corroboration of the as-built. Their absence increasingly invites an adverse inference in 2025–26 practice, on the straightforward reasoning that capturing them is now trivially cheap.
A caution on the witness statements that get written to fill the gaps. In Thomas Barnes & Sons plc v Blackburn with Darwen Borough Council [2022] EWHC 2598 (TCC), witness statements containing commentary and opinion contrary to CPR PD 57AC were heavily discounted. Delay cases are unusually exposed to this, because witness statements drift into schedule commentary almost by reflex. A statement that comments on the programme is doing an expert's job badly. The contemporaneous record is doing it properly.
What happens when the records will not support a discrete causal case
Where the records will not carry event-by-event causation, the extension of time claim becomes a global or total time claim, and that is a bad place to end. A total time claim asserts the extension as actual completion minus contractual completion, against a bundle of employer events, with no critical-path attribution. It is the weakest form of extension of time claim there is, and SCL Delay and Disruption Protocol, 2nd Edition (February 2017), Core Principle 17 discourages exactly that practice.
Walter Lilly & Co Ltd v Mackay [2012] EWHC 1773 (TCC) is the modern permissive statement and it is routinely over-read. Akenhead J held there is no rule of law preventing recovery where a contractor has not allocated loss to each event. The global approach goes to proof, not to admissibility or validity. The presence of a contractor-risk cause within the mix does not automatically defeat the claim, and a failure to keep separable records goes to credibility and weight rather than to dismissal. Permissive, but not forgiving. The claimant must still comply with conditions precedent and notice requirements, prove the events occurred and were employer risk, prove those events caused delay or loss taken as a whole, show the loss would not have been incurred anyway (for example because of an inadequate tender), and support all of it with witness and contemporaneous evidence. Every one of those obligations is a records obligation.
A note on citation discipline: the Walter Lilly global-claims synthesis is commonly pinpointed to a specific paragraph in secondary commentary, and that paragraph number has not been verified against the judgment in this corpus. Cite the case and the proposition. Do not cite a paragraph number you have not read.
The distinction pleadings keep missing is the one that matters most:
- A global claim is one where causation itself is aggregated: a number of events, no attempt to attribute a quantified part of the delay or loss to each. Its essence is causation "based substantially on inference" (London Underground Ltd v Citylink Telecommunications Ltd [2007] EWHC 1749 (TCC), Ramsey J).
- A composite claim is one where causation is individuated (each event linked to a delay or a head of loss) but quantum is aggregated because the losses genuinely cannot be dissected. A composite claim is not a global claim and is not subject to the global-claim vulnerabilities.
Failure to plead that distinction is a common and entirely avoidable pleading error, and it costs claimants the benefit of a case they had actually proved. "Global claim" is not a synonym for "big claim". It means a claim that does not link individual causes to individual effects.
If aggregation is genuinely unavoidable, the burden is to show that individuated causation is impossible rather than merely unattempted. AACE International RP 130R-23 (15 May 2023), Demonstrating Entitlement to Cumulative Impact Claims in Construction, supplies the structure for that argument (foreseeability, magnitude of change, causation, mitigation, document retention) and that distinction is the whole ballgame. One jurisdictional flag to carry with the analysis: the apportionment route identified in John Doyle Construction Ltd v Laing Management (Scotland) Ltd [2004] ScotCS 141, 2004 SC 713 is a Scots law feature and is not English law.
How this claim dies in the record set
- No accepted baseline. The employer's defence writes itself: the baseline was never accepted, so the contractor's own plan proves nothing. Impacted as-planned and time impact analysis are gone with it.
- Updates issued as PDFs. The logic, constraints and float cannot be interrogated, so no CPM-based method can be run on what survives.
- Logic changed between updates with no narrative. Adverse inferences follow: Alares Construction, Inc. v Department of Veterans Affairs, CBCA 6149, 7071, 7597 (21 March 2025).
- Progress captured as percentage complete. No defensible as-built dates, and the analysis is discounted, George Sollitt Construction Co. v United States, 64 Fed. Cl. 229 (2005).
- Delay to progress presented as delay to completion. Non-critical delay generates no extension of time however well documented it is.
- Updates recreated for the claim. AACE MIP 3.5 is expensive and invites the "reconstructed for litigation" attack, and the recreation is the first thing cross-examined.
- A global claim advanced as a first resort. It reads as a concession that the records were not kept, and tribunals read it that way.
- Witness statements doing the schedule's job. Commentary and opinion contrary to CPR PD 57AC was heavily discounted in Thomas Barnes & Sons plc v Blackburn with Darwen Borough Council [2022] EWHC 2598 (TCC).
How long do delay records now have to be kept?
In England and Wales, far longer than most retention policies assume. URS Corporation Ltd v BDW Trading Ltd [2025] UKSC 21 held that the extended limitation period introduced by the Building Safety Act 2022 s.135 (30 years retrospective and 15 years prospective for Defective Premises Act 1972 claims) has retrospective effect and applies to dependent claims, including negligence and contribution.
The delay-practice consequence is direct even though the case is about defects. Claims on projects completed decades ago are live, and where a defects claim is brought, the counterfactual programme and the completion history become relevant to quantum: what would have been built, when, and what the sequence actually was. Practitioners are already reconstructing programme records from archived projects. On a project within that exposure (dwellings, and anything where a Defective Premises Act claim or a claim dependent on one is realistic) delay and completion records should be retained against the extended horizon rather than the 12-year deed horizon. That is a project-specific judgement, not a blanket rule: set the period from the contract, the governing law, the applicable limitation regime, the defects exposure and company policy, and take advice on the project rather than inheriting a number from an article. That is a decision about archive format and media as much as about duration: native schedule files in a proprietary format that nobody can open in 2050 are not retained records, and neither are project extranets whose licence lapses at practical completion.
One boundary worth marking, because commentary crosses it freely. A related proposition circulating in 2025–26 commentary (that adjudications can now be commenced for historical defects decades after practical completion) rests on that commentary rather than on a judgment retrieved and read for this corpus. The retention consequence is stated here. The adjudication consequence is not, and should be checked against the underlying decision before anyone relies on it.
What is an AI-generated project record worth in evidence?
Nobody knows yet, and that is the honest answer. No standard-form amendment and no recognised protocol governs the evidential status of AI-generated project documentation. This should be expected to become a live authenticity and admissibility issue, and there is currently no settled practice to point to when it does.
Note the asymmetry, because it is where the gap actually sits. Instruments now address AI in expert evidence: the SVAMC Guidelines on the Use of AI in Arbitration, 1st edition (30 April 2024), Guideline 5, expressly binds experts not to use AI to falsify or compromise evidence. The Ciarb Guideline on the Use of AI in Arbitration (March 2025, updated September 2025), Article 7, requires disclosure where AI use may affect the evidence or the outcome, extends expressly to party-appointed experts at Article 7.3, and permits adverse inferences and costs consequences for non-disclosure, and the England and Wales Civil Justice Council interim report (February 2026) proposes declaration of AI use in witness statements and disclosure of substantive AI use in expert evidence: a proposal, not a rule. Nothing in that list reaches the project record itself: the AI-drafted daily report, the machine-summarised site diary, the auto-generated progress narrative.
Two related points of currency, since both are widely misreported. Proposed US Federal Rule of Evidence 707, which would have subjected machine-generated evidence offered without a supporting expert to Rule 702-equivalent reliability standards, was not adopted. The Standing Committee declined to recommend action on 3–4 June 2026 and returned the rule for revision, and there is no effective date. And the TCC Guide received a 2026 revision, its first in four years, adding sections on Building Safety Act proceedings and on the use of AI.
The defensible working position, until something governs this: where AI-assisted tools touch either the project record or the analysis, preserve the inputs and the parameters, be able to reproduce the output, and never use a tool to reconstruct as-built data the records do not support. That last one is AACE MIP 3.5 territory with an authenticity risk added on top.
The record set is a decision the project makes every week
The record set is not a state the project is in. It is a decision made every week, by whoever approves the monthly report format and decides what goes into the daily return, and the only week in which that decision can be made is the current one. Every method question in a future dispute resolves back to a set of small administrative choices taken by people who had no dispute in mind. Below is what those choices look like when they are made deliberately.
The reframing worth carrying away is that none of this is preparation for a claim. It is the ordinary administration of a project, performed slightly more deliberately, and it happens to determine which forensic methods will be available years later. The project that keeps a statused native update series with a logic-change narrative has bought itself windows analysis. The project that does not has bought itself a global claim. Neither project made that choice in a meeting about claims. Both made it in the current week, on a form template, without discussion.
Authorities
- Society of Construction Law, Delay and Disruption Protocol, 2nd Edition (February 2017) 2nd Edition (February 2017), Society of Construction LawGuidance Part B s.1 and Appendix B name the six record categories. Core Principle 1 recommends the parties agree the record regime in writing at or before the outset, Core Principle 6 confines entitlement to delay to completion, Core Principle 9 makes float identification depend on a properly prepared and regularly updated programme, Core Principle 17 discourages total time claims.Do not quote a paragraph number for the six-method table. It is variously cited between 11.5 and 11.7 across published commentary and the discrepancy is not resolved in this corpus. Separately, the finding that no 3rd edition exists rests on the absence of any 3rd edition on the SCL site and in 2025–26 commentary, and SCL International's protocols page was not retrievable. High confidence, not certainty.Protocol →
- Alares Construction, Inc. v Department of Veterans Affairs CBCA 6149, 7071, 7597 (21 March 2025), Civilian Board of Contract AppealsSchedule logic unsupported or contradicted by contemporaneous project documentation, and unexplained logic changes in the monthly updates, drew adverse inferences. 765 days claimed. 218 awarded. The current textbook illustration of what a failure to maintain the update series costs.Not a release case. The Board upheld the reservation of rights in modification P00005. The claim then failed on concurrency, an independent ground.CBCA →
- George Sollitt Construction Co. v United States 64 Fed. Cl. 229 (2005), United States Court of Federal ClaimsDelay affecting only non-critical activities does not delay completion. Separately, the expert's analysis was discounted for missing baseline schedules, absent monthly updates during startup and completion, progress reported as percentages rather than actual start and finish dates, and inaccurate critical-path activity dates.Court of Federal Claims →
- Premier Modular Ltd v Maidstone and Tunbridge Wells NHS Trust [2026] EWHC 1404 (TCC), Technology and Construction CourtAn NEC4 adjudication decision was refused enforcement in part because the adjudicator relied on an "updated" Accepted Programme when both parties agreed the programme had not been updated. Nobody can invent an update after the fact.Find Case Law →
- DCMA 14-Point Schedule Assessment: DCMA-PAM-200-1, s 4 DCMA-PAM-200-1, Earned Value Management System Program Analysis Pamphlet, s 4 (United States Defense Contract Management Agency)Fourteen structural checks on a CPM network: logic, leads, lags, relationship types, hard constraints, high float, negative float, high duration, invalid dates, resources, missed tasks, the critical path test, CPLI and BEI. The de facto first move in schedule validation.A quality screen, not a legal standard, and it should be described that way whenever it is used. Failing several checks does not make a programme non-compliant or inadmissible. Passing all fourteen does not validate it as a truthful account of the works. There is no standalone DCMA publication called the "14-Point Schedule Assessment": the checks are section 4 of DCMA-PAM-200-1, the EVMS Program Analysis Pamphlet, and should be cited that way.DCMA →
- Walsh Construction v Toronto Transit Commission 2024 ONSC 2782, Ontario Superior Court of JusticeThe owner's own consultant's finding that average RFI response time "was also excessively high" materially supported the contractor, which is what a log dated on both sides makes provable.Find on CanLII →
- Steria Ltd v Sigma Wireless Communications Ltd [2007] EWHC 3454 (TCC), Technology and Construction CourtNotice must emanate from the party claiming, so minutes recorded by a third party will not do. This is what destroys the "the employer knew anyway" fallback and why the correspondence, notices and instructions register is a proof record rather than an archive.Find Case Law →
- Thomas Barnes & Sons plc (in administration) v Blackburn with Darwen Borough Council [2022] EWHC 2598 (TCC), Technology and Construction CourtWitness statements containing commentary and opinion contrary to CPR PD 57AC were heavily discounted. Delay cases are unusually exposed to this because witness statements drift into schedule commentary by reflex.Find Case Law →
- Walter Lilly & Co Ltd v Mackay [2012] EWHC 1773 (TCC), Technology and Construction Court (Akenhead J)There is no rule of law preventing recovery where a contractor has not allocated loss to each event. The global approach goes to proof rather than to admissibility or validity. Permissive, but the claimant must still comply with conditions precedent, prove the events and their employer-risk character, prove causation taken as a whole, and support all of it with contemporaneous evidence.The global-claims synthesis is commonly pinpointed to ¶486. That paragraph number comes from secondary sources and has not been verified against the judgment in this corpus, so no paragraph number is published here, cite the case and the proposition instead.Find Case Law →
- London Underground Ltd v Citylink Telecommunications Ltd [2007] EWHC 1749 (TCC), Technology and Construction Court (Ramsey J)The essence of a global claim is causation "based substantially on inference". The test that separates a global claim from a composite claim, where causation is individuated and only quantum is aggregated.Find Case Law →
- AACE International Recommended Practice No. 130R-23, Demonstrating Entitlement to Cumulative Impact Claims in Construction 15 May 2023, AACE InternationalSupplies the structure for the argument that individuated causation is impossible rather than merely unattempted: foreseeability, magnitude of change, causation, mitigation and document retention.Table of contents →
- John Doyle Construction Ltd v Laing Management (Scotland) Ltd [2004] ScotCS 141, 2004 SC 713, Inner House of the Court of SessionThe source of the apportionment route in global claims. It is a Scots law feature and is not English law, which is the jurisdictional flag to carry with any global-claim analysis.
- URS Corporation Ltd v BDW Trading Ltd [2025] UKSC 21, UK Supreme CourtThe extended limitation period introduced by Building Safety Act 2022 s.135 (30 years retrospective and 15 years prospective for Defective Premises Act 1972 claims) has retrospective effect and applies to dependent claims, including negligence and contribution. The delay-practice consequence is a 30-year retention horizon for delay and completion records.The related proposition circulating in 2025–26 commentary (that adjudications can now be commenced for historical defects decades after practical completion) rests on that commentary, the specific judgment was not retrieved for this corpus. The retention consequence is stated here, the adjudication consequence is not.Find Case Law →
- Silicon Valley Arbitration and Mediation Center, Guidelines on the Use of AI in Arbitration 1st Edition (30 April 2024)Guideline 5 expressly binds experts not to use AI to falsify or compromise evidence. It reaches expert evidence, not the project record itself.Guidelines →
- Chartered Institute of Arbitrators, Guideline on the Use of AI in Arbitration March 2025, updated September 2025Article 7 requires disclosure where AI use may affect the evidence or the outcome, extends expressly to party-appointed experts at Article 7.3, and permits adverse inferences and costs consequences for non-disclosure. Again expert evidence, not the project record.Guideline →
About this material Library content is general information about construction claim practice, not legal advice. Entitlement, deadlines and procedure are governed by your own contract and by the law of the place the work is performed. Reviewed August 7, 2026 · Report a correction
In this article
- Which delay analysis methods have your records already ruled out?
- Why the schedule update series is the load-bearing record
- What schedule validation actually checks, and what it does not
- Daily reports, manpower and when the impact actually began
- What happens when the records will not support a discrete causal case
- How long do delay records now have to be kept?
- What is an AI-generated project record worth in evidence?
- The record set is a decision the project makes every week