Walk into any turnaround readiness pitch and you’ll hear a version of the same statistic: 80% of turnarounds miss their budget or schedule. Sometimes it’s 82%. Sometimes it’s “most turnarounds fail.” The number moves around because almost nobody quoting it knows where it came from.
This page traces the claim to its sources, scores each version against the Reliable Confidence Score, and lays out the turnaround overrun figures that a refinery or petrochemical audience can cite without getting torn apart.
What These Statistics Measure, and Where They Come From
California Labor Code Sections 7872-7873, the statutory definition Cal/OSHA works from, define a turnaround as a planned, periodic shutdown, total or partial, of a refinery process unit or plant to perform maintenance, overhaul, and repair operations and to inspect, test, and replace process materials and equipment.
Similar language is often attributed to the American Petroleum Institute, though the original API page is not readily locatable. Lawrence, writing for AP-Networks in PTQ, puts the typical cycle at four to six years and the typical execution window at three to five weeks. AP-Networks describes the largest events as involving more than a million labor hours.
Turnaround overrun statistics come from three kinds of sources, and they aren’t equally solid.
The first is proprietary benchmarking databases. Asset Performance Networks (AP-Networks) maintains a turnaround database it reported at over 1,350 turnaround observations in a published whitepaper. Independent Project Analysis (IPA) maintains capital project databases exceeding 20,000 projects, including turnaround-related work. Both firms publish selected findings; the underlying data stays private.
The second is published samples with named authors and defined methods. The strongest example is a cost estimating analysis by Gordon Lawrence of AP-Networks, published in PTQ and available on DigitalRefining, built on a defined sample of 93 turnarounds.
The third is consultant lore. The famous 80% and 82% figures belong here. Vendor and trade pages attribute them to T.A. Cook, a Berlin-based maintenance consultancy acquired by Accenture in November 2021, and in at least one case to T.A. Cook and Solomon Associates jointly, dated 2019. We could not locate the original study, a sample size, or a methodology in any open source. What circulates is the attribution, not the research.
The Reliable Confidence Score Table
| Claim | Figure | Reliable Confidence | What It Really Means |
|---|---|---|---|
| “80% of turnarounds exceed budget” (as commonly quoted, no source) | 80% | Lowno traceable source | The version that circulates in vendor blogs and pitch decks, with attribution stripped. Vendor sites cite each other, not a study. Don’t quote it bare. |
| T.A. Cook / Solomon-attributed cluster (2019): 82% miss performance expectations, ~50% delayed, 80% over budget by more than 10% | 82% / 50% / 80% | Lowconsistent attribution, untraceable study | Accruent credits the figures to T.A. Cook and Solomon Associates (2019); ISHN credits a similar cluster to T.A. Cook alone. The underlying study, sample, and definitions are not publicly available. Cite as “attributed to,” never as a verified finding. |
| More than two-thirds of turnarounds exceed planned cost and schedule by 10%, or have a trip after startup | >67% | Highpublished AP-Networks finding | Quoted exactly as published (Hansen and Schroeder, AP-Networks whitepaper, 1,350+ turnaround database): “exceed their planned cost and schedule by 10 percent or have a trip after startup.” Keep that structure intact; the trip clause and the threshold get dropped in almost every requote. |
| 40% of turnarounds have a cost overrun or schedule delay of more than 30% | 40% | Highpublished AP-Networks finding | Same whitepaper. This is the severe-failure rate, and it’s the more useful number: nearly half of turnarounds blow through their targets by nearly a third. |
| Turnaround budgets underestimate actual cost by roughly 16% on average | ~16% | Highpublished sample of 93 turnarounds | Lawrence (AP-Networks) compared actual cost against control budget for 93 turnarounds. Budgets were intended as ±10% estimates by 83% of practitioners surveyed; results behaved more like ±30% estimates. |
| Almost 50% of turnarounds change their start date at least once | ~50% | Mediumstated on AP-Networks’ own pages, no published methodology | AP-Networks reports this as a leading indicator of failure. Directionally credible from the database owner, but the supporting analysis isn’t published. |
| Less than 45% of capital projects finish within ±10% of their authorization estimate; average execution schedule slip is 26% | <45% / 26% | Mediumstated on IPA’s own site; page not accessible to automated verification | IPA capital project figures, not turnaround-specific. The CSRA page serves a JavaScript gate, so the figures rest on captures of the rendered page rather than a directly retrievable document. Useful context; don’t relabel as turnaround statistics. |
“80% of turnarounds exceed budget” (as commonly quoted, no source)
Figure80%
Reliable ConfidenceLowno traceable source
What It Really MeansThe version that circulates in vendor blogs and pitch decks, with attribution stripped. Vendor sites cite each other, not a study. Don’t quote it bare.
T.A. Cook / Solomon-attributed cluster (2019): 82% miss performance expectations, ~50% delayed, 80% over budget by more than 10%
Figure82% / 50% / 80%
Reliable ConfidenceLowconsistent attribution, untraceable study
What It Really MeansAccruent credits the figures to T.A. Cook and Solomon Associates (2019); ISHN credits a similar cluster to T.A. Cook alone. The underlying study, sample, and definitions are not publicly available. Cite as “attributed to,” never as a verified finding.
More than two-thirds of turnarounds exceed planned cost and schedule by 10%, or have a trip after startup
Figure>67%
Reliable ConfidenceHighpublished AP-Networks finding
What It Really MeansQuoted exactly as published (Hansen and Schroeder, AP-Networks whitepaper, 1,350+ turnaround database): “exceed their planned cost and schedule by 10 percent or have a trip after startup.” Keep that structure intact; the trip clause and the threshold get dropped in almost every requote.
40% of turnarounds have a cost overrun or schedule delay of more than 30%
Figure40%
Reliable ConfidenceHighpublished AP-Networks finding
What It Really MeansSame whitepaper. This is the severe-failure rate, and it’s the more useful number: nearly half of turnarounds blow through their targets by nearly a third.
Turnaround budgets underestimate actual cost by roughly 16% on average
Figure~16%
Reliable ConfidenceHighpublished sample of 93 turnarounds
What It Really MeansLawrence (AP-Networks) compared actual cost against control budget for 93 turnarounds. Budgets were intended as ±10% estimates by 83% of practitioners surveyed; results behaved more like ±30% estimates.
Almost 50% of turnarounds change their start date at least once
Figure~50%
Reliable ConfidenceMediumstated on AP-Networks’ own pages, no published methodology
What It Really MeansAP-Networks reports this as a leading indicator of failure. Directionally credible from the database owner, but the supporting analysis isn’t published.
Less than 45% of capital projects finish within ±10% of their authorization estimate; average execution schedule slip is 26%
Figure<45% / 26%
Reliable ConfidenceMediumstated on IPA’s own site; page not accessible to automated verification
What It Really MeansIPA capital project figures, not turnaround-specific. The CSRA page serves a JavaScript gate, so the figures rest on captures of the rendered page rather than a directly retrievable document. Useful context; don’t relabel as turnaround statistics.
The Big Takeaway
The 80% claim isn’t invented out of thin air. It’s an orphaned statistic. The T.A. Cook attribution appears in multiple vendor and trade sources, which suggests a real study once existed. But a figure whose methodology, sample, and definitions can’t be inspected can’t be defended, and in front of a turnaround steering committee it will get challenged.
The defensible story is nearly as dramatic and far better sourced. AP-Networks’ published data puts more than two-thirds of turnarounds over their cost and schedule targets by 10% or worse (or tripping after startup), and 40% over by more than 30%.
Nearly half of turnarounds don’t just miss their targets. They miss them by almost a third. That’s the number worth putting in front of a steering committee.
The Lawrence sample sharpens the point. Turnaround teams tell themselves their control budgets are ±10% estimates. Measured against 93 actual outcomes, the budgets behaved like ±30% estimates, with a mean cost underestimate around 16%. The industry’s problem isn’t only execution. It’s calibration.
Why the Numbers Vary and Disagree
The circulating figures disagree because they measure different things against different thresholds.
“Miss budget” is meaningless without a tolerance. A turnaround 3% over budget technically missed, but no owner treats that as failure. AP-Networks uses explicit thresholds (10% and 30%), which is why its two figures look so different from each other and from the flat “80%.”
Composite measures inflate quietly. The AP-Networks two-thirds figure covers turnarounds that exceed planned cost and schedule by 10% or have a trip after startup. Quote only one component and you’re describing a different number, one AP-Networks hasn’t published. Requoters keep the big figure and drop the definition.
Scope keeps moving. AP-Networks reports industry average scope growth of 19% between scope freeze and execution, against 7% for top quartile performers. When scope grows a fifth after the budget locks, the “overrun” partly measures scope discipline, not estimating error.
Samples differ. AP-Networks does publish partial composition for its database: 1,350+ turnarounds spanning onshore and offshore upstream, gas processing, refining, chemicals, and power generation, with over 60% rated High to Mega complexity on its four-tier scale. That skew toward large, complex events matters. A database heavy on megaturnarounds will show different outcomes than a mix of small chemical plant outages, and no proprietary database publishes enough composition detail to reweight its findings for your specific plant type.
How to Use These Statistics Safely
Cite the AP-Networks figures with their full definitions, including the 10% threshold and the trip clause. The precision is what makes them credible.
Use the 40%-over-30% figure when you need to make the severity case. It’s better sourced than the 80% claim and lands harder, because it describes catastrophic misses rather than technical ones.
If you must reference the 80% or 82% figures, attribute them explicitly: “figures attributed to T.A. Cook and Solomon Associates, though the original study is not publicly available.” That phrasing survives scrutiny. A bare “80% of turnarounds fail” does not.
Keep capital project statistics labeled as capital project statistics. IPA’s slip and estimate accuracy figures are strong, but they describe a different population.
And connect overrun risk to what the downtime itself costs. A turnaround that slips two weeks is buying two extra weeks of lost production on top of the labor overrun. Our industrial downtime cost benchmarks cover the per-hour side, and our unplanned downtime frequency benchmarks cover how often the unplanned kind hits.
Where Teams Go Wrong
The most common error is quoting the 80% figure as settled fact. Vendor and trade pages repeat the figures, but the trail stops at a T.A. Cook or T.A. Cook and Solomon Associates attribution rather than an inspectable study. Accruent credits “TA Cook and Solomon Associates (2019)” with no link to the research; ISHN credits T.A. Cook alone; one STO software vendor cites a competing vendor’s blog as its source. None of the trails reach a study.
The second error is threshold-stripping: quoting “two-thirds exceed cost and schedule” without the 10% qualifier or the startup-trip clause, which turns a defined composite metric into a vague one.
Estimating discipline is part of the same story. In a survey Lawrence reports in the PTQ article, only 34 of 216 turnarounds (16%) could clearly separate emerging work, discovery work, and contingency in their cost estimates. Teams that can’t tell those three apart can’t learn from their own overruns.
The third is treating overrun statistics as fixed laws rather than distributions. AP-Networks’ own published data shows top quartile performers holding scope growth to 7% against a 19% industry average. The spread between average and good is the whole argument for front-end loading, scope freeze discipline, and stable start dates. IPA reports that more than half of sites also fail to integrate turnarounds with capital projects executed in the same window, which is one of the identifiable mechanisms behind the misses.
Methodology
We searched for the primary source behind the widely quoted 80% and 82% turnaround failure figures across trade press, consultant publications, and vendor content. In the public examples we found, the figures are attributed to T.A. Cook, and in one case to T.A. Cook and Solomon Associates jointly (2019), but no public study, sample, or methodology could be located. T.A. Cook was acquired by Accenture in November 2021, per Accenture’s newsroom. We rated those figures Low and documented examples of attribution-stripped circulation.
We anchored the defensible figures to two published AP-Networks sources: the whitepaper “Benchmarking and Optimizing Maintenance Work Scope for Turnarounds” (thresholds, database size, and scope growth figures stated in the document) and Gordon Lawrence’s PTQ article “Cost estimating for turnarounds” (defined 93-turnaround sample).
IPA figures come from IPA’s own published pages and are labeled as capital project statistics; because ipaglobal.com serves a JavaScript gate to automated retrieval, the CSRA figures are rated Medium. All other source URLs were confirmed live at publication. Confidence ratings reflect source traceability and fitness for the specific claim, not how often a figure is quoted.
Frequently Asked Questions
Do 80% of turnarounds really go over budget?
The 80% figure is widely attributed to T.A. Cook research, but the original study is not publicly available and no methodology can be inspected. The best published alternative comes from AP-Networks: more than two-thirds of turnarounds exceed planned cost and schedule by 10% or have a trip after startup, and 40% overrun cost or schedule by more than 30%.
What percentage of turnarounds finish on time and on budget?
Based on AP-Networks’ published whitepaper data, fewer than one-third of turnarounds stay within 10% of their cost and schedule targets without a post-startup trip. No published open dataset reports an exact “fully successful” percentage with an inspectable methodology.
What is the average turnaround cost overrun?
A published AP-Networks analysis by Gordon Lawrence, using a sample of 93 turnarounds, found budgets underestimated actual costs by roughly 16% on average, even though most practitioners intended their budgets as ±10% estimates.
Where does the “82% of turnarounds fail” statistic come from?
Vendor and trade pages attribute it to T.A. Cook, a maintenance consultancy acquired by Accenture in November 2021, and in at least one case to T.A. Cook and Solomon Associates jointly (2019). The cluster runs: 82% miss performance expectations, about half experience delays, and 80% run more than 10% over budget. The underlying study has never surfaced publicly, so the figures can’t be verified.
How much does turnaround scope grow after scope freeze?
AP-Networks’ published whitepaper reports industry average scope growth of 19% between scope freeze and execution, based on its database of more than 1,350 turnarounds. Top quartile performers hold scope growth to about 7%.
Why do turnarounds overrun so often?
Published sources point to a few repeat offenders: budgets set as ±10% estimates that behave like ±30% estimates (Lawrence, AP-Networks), scope growth after freeze, changed start dates (AP-Networks reports nearly half of turnarounds move their start date at least once), and poor integration between turnaround work and capital projects executed in the same window (IPA).
Are turnaround overruns worse than capital project overruns?
They’re comparable problems measured differently. IPA states on its site that less than 45% of capital projects finish within ±10% of their authorization estimate and that the average project slips its execution schedule by 26%. Turnaround-specific published data (AP-Networks) shows more than two-thirds missing a 10% cost-and-schedule threshold or tripping at startup.
The Short Version
The “80% of turnarounds miss budget or schedule” claim traces to T.A. Cook attributions with no locatable study behind them, so it rates Low. The citable versions come from AP-Networks’ published data: more than two-thirds of turnarounds exceed cost and schedule targets by 10% or trip after startup, 40% overrun by more than 30%, and a 93-turnaround sample shows budgets underestimating actual costs by about 16% on average. Quote the defined figures with their thresholds, attribute the folklore honestly when you must use it, and keep capital project statistics out of the turnaround column.
Sources
Shawn Hansen and Brett Schroeder (Asset Performance Networks), “Benchmarking and Optimizing Maintenance Work Scope for Turnarounds” (whitepaper):https://www.ap-networks.com/wp-content/uploads/2021/11/WhitePaper-Benchmarking-and-Optimizing-Maintenance-Work-Scope-for-Turnarounds.pdf
AP-Networks, “Optimization for Timetables of Turnarounds”:https://www.ap-networks.com/optimization-for-timetables-of-turnarounds/
Gordon Lawrence (Asset Performance Networks), “Cost estimating for turnarounds,” PTQ / DigitalRefining: https://www.digitalrefining.com/article/1000335/cost-estimating-for-turnarounds
Lawrence article PDF mirror (contains full figures): https://www.stocontrol.com/wp-content/uploads/2019/06/Cost_Estimating_For_Turnarounds-2.pdf
Independent Project Analysis, Cost & Schedule Risk Analysis (CSRA): https://www.ipaglobal.com/services/cost-engineering/cost-schedule-risk-analysis-csra/
Independent Project Analysis, “Successful Integration of Turnarounds and Capital Projects” (webinar page): https://www.ipaglobal.com/webinar/successful-integration-of-turnarounds-and-capital-projects/
ISHN, “Transform turnarounds” (T.A. Cook attribution example): https://www.ishn.com/articles/113307-transform-turnarounds
Accruent, “Understanding 5 Phases of a Plant Turnaround Process” (T.A. Cook / Solomon Associates 2019 attribution example): https://www.accruent.com/resources/blog-posts/understanding-5-phases-plant-turnaround-process
Cal/OSHA Process Safety Management Unit (turnaround definition, Labor Code Sections 7872-7873): https://www.dir.ca.gov/dosh/psm-unit.html
Accenture newsroom, “Accenture Acquires T.A. Cook” (November 15, 2021): https://newsroom.accenture.com/news/2021/accenture-acquires-ta-cook-strengthening-asset-performance-management-capabilities-for-clients-in-capital-intensive-industries
