2026 Headline Numbers
Disputes aren’t a paperwork problem, they’re a material cost risk. HKA’s CRUX Insight report (Nov 2025) covers 2,200+ major projects in 114 countries. On distressed jobs, the average sum in dispute is 33.4% of the contract budget and the average time claimed is 65.8% of planned duration. Treat that as an exposure you control early, not a fight you “win” at the end.
Most programmes fail because the baseline is weak on day one. A 2025 benchmark review of 70,000+ CPM programmes found 76% finished later than the original baseline end date. Only 12% of baseline schedules met high-quality standards. If the logic, links, and constraints are poor at award, the site team spends the job reacting, then arguing about EOT instead of managing sequence.
“Within 10% of budget” is a high bar, not a normal outcome. KPMG’s global construction survey, cited in 2026 industry briefings, puts the share of projects landing within 10% of the original budget at 25%. If your tenders assume tight delivery as the default, your margin is already spent before the first valuation goes in.
| Headline metric (plain definition) | Benchmark | Source |
|---|---|---|
| Average sum in dispute on distressed projects (as % of contract budget) | 33.4% | HKA CRUX Insight, Nov 2025 |
| Average time claimed on distressed projects (as % of planned duration) | 65.8% | HKA CRUX Insight, Nov 2025 |
| Projects finishing late vs baseline end date (original approved programme) | 76% late | State of Construction Scheduling, 2025 (70,000+ CPM programmes) |
| Baseline schedules meeting high-quality standards (at day one) | 12% | State of Construction Scheduling, 2025 (70,000+ CPM programmes) |
| Projects finishing within 10% of original budget | 25% | KPMG global construction survey (cited in 2026 industry briefings) |
Weekly beats monthly. If you wait for the month-end CVR to see programme slip and change drift, you find it when it’s too late to protect margin. Archdesk teams that hold one weekly cost-to-complete check and one change-control review per project catch problems while they are still small enough to fix.
Where Overruns Cluster
Overruns cluster around two things you can spot early, long-lead packages and slow decisions. UK delivery is the clearest example. Elecosoft’s 2025 review reports 95% of projects are delayed, with a median delay beyond 200 days. That’s not a planning issue. It’s a pricing and prelims risk, because site overhead, labour standing time, and re-sequencing become your biggest cost line when the programme stretches.
Middle East overruns often come from admin drag sitting on top of inflation. A 2025 Middle East market review cited by DynamicsSmartz reports 83% average delay versus schedule. The same review cites Turner and Townsend’s 2026 outlook of 5% to 7% building cost inflation in Saudi and 3% to 5% in the UAE. Add the compliance load in that review, about 17% of total project cost, and late submittals start costing more than late concrete.
| Hotspot | What clusters there | Indicator | What to do in the bid and the job | Source |
|---|---|---|---|---|
| UK | Programme stretch | 95% delayed, median delay beyond 200 days | Price prelims for reality. Tie client dates to procurement and access, not “target” milestones. | Elecosoft, 2025 |
| Saudi and UAE | Approvals plus inflation | 83% avg delay. 5% to 7% inflation (KSA), 3% to 5% (UAE). Compliance about 17% of project cost. | Run an approvals tracker like a programme. Escalate late sign-offs the same way you escalate late materials. | Regional market review, 2025-26 (cited by DynamicsSmartz); Turner and Townsend, 2026 outlook (cited) |
| US | Procurement volatility | Inputs up 41.6% from Feb 2020 to Mar 2025. Tariffs affected about 70% of contractors. Cable prices up 22% year on year. | Shorten quote validity and reprice long-leads at order. Don’t carry tariff risk in a fixed rate. | ABC data reported by Newsweek (to Mar 2025); WSINC update (tariffs and cable pricing) |
North America’s hotspot is procurement, not production. Associated Builders and Contractors data reported by Newsweek shows US input prices rose 41.6% from Feb 2020 to Mar 2025. WSINC reports tariffs have affected about 70% of contractors, with a 4% to 8% total cost impact depending on exposure. Treat that as a commercial control, not an estimating footnote. If your buyout happens late, your margin is already gone.
Run a simple hotspot dashboard per job: late materials percent, approvals aging, and value of pending changes. Archdesk users who review those three numbers weekly spot the cluster early enough to reset procurement dates, push decisions up the chain, and protect prelims. Build it into your project reviews, not your claims process.
Root Causes Ranked
Scope change is the biggest trigger, but slow approval is what burns margin. A 2023 Arcadis disputes analysis found scope change appeared in 73% of major disputes. The cost jumps when the site team keeps moving and the change sits waiting for instruction, because prelims and disruption keep running.
Schedule breakdown shows up first as churn, not as a missed finish date. CII schedule performance research links poor outcomes to unstable critical paths and out-of-sequence progress. Track weekly critical activities that drop off the list, because that is the early warning that trade sequencing is already failing.
Estimating error is where you lose the job before you start it. One benchmark cited in the drafts puts the average gap between the initial estimate and the final trade package award at 14%. That gap turns into under-allowed hours, then turns into variation noise and a tougher commercial relationship on every valuation.
Materials delay hurts most when it forces re-visits, not when it raises purchase price. Dodge Construction Network’s 2024 contractor survey found 57% had at least one project delayed by material lead times in the prior year. Late long-lead kit drives temporary works, partial completions, and return trips, and each return trip drags output down for every trade behind it.
| Root cause | What it looks like early | Weekly metric that catches it | Source |
|---|---|---|---|
| Scope change | Work starts without instruction, then gets re-sequenced | Days from change request to instruction | Arcadis disputes analysis, 2023 |
| Schedule churn | Critical path keeps moving, out-of-sequence working increases | Count of critical activities replaced week to week | CII schedule performance research, 2023-2024 reviews |
| Estimating gap | Trade packages land above tender allowance | % gap, allowance vs awarded value per package | Industry benchmarking cited in drafts |
| Lead-time disruption | Temporary works and partial completions increase | Re-visit count per workface, per week | Dodge Construction Network contractor survey, 2024 |
Most jobs don’t “blow up” in one moment. They drift through small, slow decisions that force re-sequencing, then the cost shows up weeks later in labour hours and prelims.
Run three gates every week and treat misses as commercial risk, not site noise: change request to instruction days, RFI ageing, and planned versus actual percent complete by workface. Archdesk is built to keep those three numbers live across site and commercial, so you spot drift early enough to reset the plan and price the change properly.
The True Cost
Delay kills margin through time-related cost you often can’t claim back. On a typical £500k package, every extra week keeps supervision, access, plant and temp services running, but the value side barely moves. Track it as a separate cost bucket and you’ll see margin fade early, not at final account.
Rework and waiting time look small on a day sheet, then show up as a lost month at the end. One 2025 industry prevention guide puts commercial rework at 5% to 10% of total labour cost. Put a simple number against “no progress” time as well. A four-person crew held for two hours, three times a week, burns about $1,080 a week at $45 an hour. That is cash you won’t recover unless you notice it while you still have notice windows open.
| Cost you feel first | What causes it on site | What it does to the business | Evidence in drafts |
|---|---|---|---|
| Extended prelims | Extra supervision, access gear, temp services, welfare | Margin fade and weaker cash cover for the next job | 2026 preconstruction note referenced in drafts |
| Rework | Fixing and repeating work after late design info or trade clashes | Labour hours inflate with no clean claim route | Industry prevention guide, 2025: 5%–10% of labour cost |
| Idle time | Waiting on access, permits, late materials, trade stacking | You pay wages now, then fight for recovery later | Crew example in drafts: $45/hr × 4 × 2 hrs × 3/wk = $1,080/wk |
| Overtime premium | Trying to “catch up” after lost windows | Costs rise faster than progress, especially in MEP and fit-out | Industry outlook, 2026: overtime 20%–30% above base wage |
The turning point isn’t practical completion. It’s the first valuation cycle where committed cost rises faster than certified value. Catch that week and you still have time to issue notices, reset the forecast, and stop “working it out” on site.
Run a simple control next week on your three most stretched jobs. Update cost-to-complete every week, not monthly. Split time-related cost into four lines: supervision, access, plant, and temp services. Archdesk teams see the same pattern repeatedly: once those four lines are visible, delivery teams stop carrying delay quietly and the commercial team gets the facts early enough to protect margin.
Who Gets Hit
Megaproject packages carry a different risk class, even if your own scope looks clean. Bent Flyvbjerg’s Oxford database of 16,000 plus projects across 136 countries reports a mean cost overrun of 62%, with rail averaging 44.7% (cited in 2026 infrastructure briefings). Big jobs fail for a simple reason: the interface load explodes, then small coordination misses turn into programme and cost events.
Public infrastructure gets hit early by estimating and approval drag, not just bad site delivery. The National Cooperative Highway Research Program (NCHRP) links estimating errors to 32% of highway cost overruns. That matters to specialists too, because tender allowances get locked in before the design is stable, then you spend the job arguing what was and wasn’t included.
| Project profile | Where margin usually leaks | Weekly metric that spots it first | Source |
|---|---|---|---|
| Megaproject packages | Interface clashes, re-sequencing, access constraints | Out-of-sequence tasks logged vs plan | Flyvbjerg Oxford database, cited 2026 |
| Public highways and civils | Tender allowance gaps and slow instructions | Allowance vs award gap per trade package | NCHRP research cited in drafts |
| Retrofit and live environments | Unknown conditions, rework, short-notice change | RFIs per £1m plus rework hours booked | 2026 estimating and pricing outlooks cited in drafts |
Retrofit is volatile because you can’t see the work at tender stage. One 2026 estimating review cited in the drafts warns estimating gaps can run up to 20% of total budget on refurbishment work. That gap rarely comes back as a clean variation. It shows up as rework and stop-start labour, so the cost sits in wages and prelims, not in materials.
Jobs don’t hurt because they’re big. They hurt because the interface count is high and the scope is still moving after award. Treat that as a pricing and management choice, not bad luck.
Run a quick risk score before you bid and again at start. Count live interfaces, test design maturity, and time how long it takes to turn a query into an instruction. Track those three weekly in Archdesk alongside cost-to-complete. The early drift shows up in handovers and rework hours weeks before it hits the headline margin.
Tech and Outcomes
Tech only improves outcomes when it cuts decision latency. Most firms can capture labour time, photos, and site diaries. The margin gain comes when that data reaches a PM, QS, or buyer fast enough to change tomorrow’s plan, not next month’s report. Archdesk sees the same pattern across live jobs. Weekly control beats “more dashboards” every time.
Weekly cost-to-complete works because it forces one hard conversation early. The question isn’t “are we 60% complete”. It’s “does measured production match what we’ve spent and what we’ve still committed”. A 2026 scheduling and AI management review reports contractors using advanced scheduling cut planning and reporting time by 15% to 20%. Time saved matters, but the bigger win is fewer arguments about percent complete because progress is tied back to cost codes and earned value.
Spreadsheets keep teams “fast”, but the numbers aren’t trusted. Audit work cited in the drafts repeatedly puts spreadsheet error rates at 88% to 90%. That drives a real site behaviour. People stop acting on the cost position because they don’t believe it, so decisions slide to month-end. A system is only live when you can see committed cost, change exposure, and procurement status without chasing emails.
| Control step | What you standardise on site | What it changes commercially | Evidence |
|---|---|---|---|
| Daily capture | Labour time and plant time coded to the job cost structure | Less “unknown” cost sitting in gangs and prelims | Industry reports referenced in drafts, 2025–2026 |
| Single cost view | Budget, committed cost, and forecast in one place, not emailed versions | Fewer late shocks at application time | Spreadsheet error rate 88%–90% cited by audit work in drafts |
| Weekly CTC | Measured progress by cost code, matched to what’s spent and bought | Earlier replan on labour, buying, and subcontract scope | 2026 scheduling and AI management review: 15%–20% admin time cut |
Practical move: pick one live job and run a four-week weekly CTC routine with named owners. Lock three inputs, daily labour time, weekly measured progress by cost code, and a live log of committed procurement. Use Archdesk to keep those three feeds in the same place, then force one decision each week: re-sequence, re-resource, buy earlier, or raise the change. The goal is simple. Make the forecast change faster than the job changes.
What Leaders Do
Leaders win margin by shortening the time between a warning sign on site and a commercial decision. The practical difference is cadence. Leaders run the key control loops weekly, with named owners and clear thresholds. Laggards find out at month end, when the only options left are overtime, claims, or write-off.
Weekly Earned Value stops the “percent complete” argument before it starts. CPI is Earned Value divided by Actual Cost. A CPI below 1.0 means you are overspending on the plan you agreed. Leaders track CPI weekly at work-package level and force a re-plan when the variance is more than 5% to 10% of that package budget, not when the CVR lands.
Leaders don’t start packages “to keep busy”. They start when design is issue-ready and buy-out is dated. That one gate removes the most common cause of false progress: labour on site with no installed value. Make “buy-out complete” a pass/fail check with need dates, not a dashboard target.
Change control is a production rule, not a QS rule. Leaders stop out-of-sequence work that lands without a signed instruction. The commercial reason is simple: once the work is buried, you lose the clean link between cause, time, and cost. Track “days from change request to signed instruction” weekly. If it stretches, programme and cash stretch with it.
| Control loop | Leader habit (weekly) | Metric that triggers action | What it protects |
|---|---|---|---|
| Estimating | Review allowances against live package returns, not just at tender review | Allowance vs award gap per package | Next bid margin and current buy-out risk |
| Procurement | Dated procurement schedule with need dates agreed to the programme | % committed vs need date | Sequence and prelims exposure |
| Production planning | 2–6 week lookahead with a constraint removal log | PPC (Percent Plan Complete) | Labour productivity and trade handovers |
| Change control | Signed instruction before work starts, with a dated log | Days from request to instruction | Variation recovery and delay ownership |
| Commercial | Cost-to-complete reforecast with PM and QS in the same room | CPI trend (EV ÷ AC) | Margin fade and cash predictability |
Practical move for the next 30 days: run a weekly “control room” on one live job, with five numbers only. Use commitment vs need date, PPC, open constraints, change ageing, and CPI. Keep it to 30 minutes. If you can’t produce those numbers quickly and consistently, fix the data flow first. Archdesk clients see the fastest lift when budget, committed cost, and forecast sit in one live view that site, commercial, and buying can all trust.
2026 Outlook Signals
Inflation headlines are calmer, but delivery risk is harder to control. Turner and Townsend’s 2025 market review flags stabilising cost inflation in many markets, while labour immobility and skills gaps stay stubborn. That’s a commercial problem, not a macro one. You can win a tender on price and still lose margin if you can’t lock gangs and long-leads to the programme early.
Labour is still the baseline constraint, and the hidden cost is lost output. AGC’s 2025 workforce survey says 92% of firms that are hiring struggle to find people. The same survey says 57% of candidates are not qualified. Rate rises hurt, but productivity loss hurts faster. Gang churn drives re-sequencing, supervision drag, and more defects that never turn into clean variations.
| Signal to track | What it usually turns into | What to do early | Source |
|---|---|---|---|
| Trade continuity, gang turnover by package | Programme churn and rework hours | Lock gangs to work areas and pay for continuity, not just attendance | CIC Construction, 2026 (cited in drafts) |
| Project abandonment frequency in your client segment | Suspensions, scope trims, delayed payments | Audit funding and approvals before order, and cap exposure on prelim-heavy scopes | Deloitte, 2025 (cited in drafts) |
| Late payment incidence, upstream and downstream | Supplier slowdown and stoppages | Pre-agree vesting and step-in rights for key materials, and enforce interest | Payments report, 2026 (cited in drafts) |
| Tariff or trade-policy shocks on imported packages | Repricing, lead-time slips, redesign | Buy early on long-leads and write price adjustment terms into orders | Deloitte, 2025; OVABC, 2026 (cited in drafts) |
The base-case risk for 2026 to 2027 is “death by drift”. Jobs don’t fail on tender price. They fail on labour continuity, long-lead timing, and slow commercial decisions.
Price your control, not your optimism. Run three tender scenarios and set terms for the downside. Push hard on dated need-by commitments, clear suspension language, and fast instruction turnarounds. Archdesk sees the same pattern across live projects. Firms that forecast weekly and close change fast hold margin even when the market turns mid-job.
Frequently Asked Questions
What percentage of construction projects are delayed in 2026?
Elecosoft's 2025 UK review reports 95% of projects are delayed, with a median delay beyond 200 days. Globally, HKA's CRUX Insight report across 2,200 plus distressed projects shows the average time claimed in disputes is 65.8% of planned duration. These aren't outliers. They reflect a structural delivery problem across regions and project types.
What is the biggest cause of construction cost overruns?
Scope change is the top trigger, appearing in 73% of major disputes according to Arcadis's 2023 analysis. The real cost driver isn't the change itself, it's the delay in getting instructions while prelims and disruption keep running. Fast approval processes matter more than tighter scope definitions.
How much margin does a one-week delay actually cost on a subcontract package?
On a typical £500k package with a planned 20% gross margin, every extra week keeps supervision, plant, access, and temp services running while earned value barely moves. Track these time-related costs as a separate bucket from day one. If you wait until final account to quantify them, you've already lost the margin and the leverage to recover it.
Why are megaprojects more likely to overrun than mid-market jobs?
Bent Flyvbjerg's Oxford database of 16,000 plus projects across 136 countries shows a mean cost overrun of 62%, with rail averaging 44.7%. The cause is interface load. On megaprojects, the number of coordination handoffs between trades and packages grows exponentially, so small misses compound into programme-level failures that hit every subcontractor on site.
Does construction management software actually reduce project delays?
Software only improves outcomes when it cuts the time between a site problem and a commercial decision. Firms that track cost-to-complete weekly finish closer to budget than those reporting monthly, because site teams stop hiding problems when they know issues will surface in days. More dashboards don't help. Faster decisions do.
What do high-performing contractors do differently to stay on budget?
They run weekly control loops with named owners and clear cost thresholds, rather than waiting for month-end reports. Weekly Earned Value reviews kill the "percent complete" argument early, and weekly cost-to-complete forces one honest conversation about where the job really stands. The practical difference is cadence, not cleverness.
How large are disputed sums on distressed construction projects?
HKA's CRUX Insight report covering 2,200 plus projects in 114 countries puts the average sum in dispute at 33.4% of the contract budget. That's not a paperwork problem, it's a material cost exposure. Contractors who treat dispute risk as something to manage from week one, through contemporaneous records and clear variation processes, protect far more margin than those who try to "win" at final account.
What are the key delivery risks for construction contractors through 2026?
Turner and Townsend's 2025 market review flags stabilising cost inflation, but labour immobility and skills gaps remain stubborn. You can win a tender on price and still lose margin if you can't lock gangs and long-lead materials to the programme early. The risk has shifted from headline material costs to programme control and resource availability.
