Methodology
The methodology behind the first credible UK NEC outcomes benchmark.
Eight rules. Contributor consent. A k-anonymity floor of ten. Methodology published nine months before the data, in the open, for critique.
Method preview · v0.1 · open for critique
Why methodology-led
The data isn’t the moat. The methodology is.
Anyone with enough clients can publish a survey. Few publishers on UK NEC have actually published one — and the reason is that the surveys that exist do not survive serious scrutiny. Definitions drift. Sample frames are unstated. The interesting cells are below the floor at which they would be safe to publish, but get published anyway. The aggregate looks credible until someone presses on it.
The Benchmark is built on the opposite principle. The method ships before the data. The method is published in full so it can be argued with. The data, when it comes, is anchored to a method that is already in the open. That is how a reference set becomes citable.
The eight rules, expanded
Each rule, what it means, and why it can't bend.
01
Contributor consent
Data only enters the benchmark via signed consent from the contributing party. No scraping. No reconstruction.
Each contributing party signs a published consent statement before any contract data is shared. The statement specifies what data will be collected, how it will be anonymised, what cells it will roll into, and the contributor's right to withdraw. Withdrawal removes the contract from forward issues but does not retroactively change a published issue (see rule 07).
02
k-anonymity ≥ 10
No aggregate is published unless at least ten distinct contracts contribute to the cell. No exceptions.
Every cell of every published table — sector × Option × metric — must clear a floor of ten distinct contributing contracts before publication. Cells that do not clear the floor are reported as 'below k=10 floor; not published'. The floor is non-negotiable; reducing it reduces contributor protection, which collapses the participation incentive that makes the benchmark possible.
03
Sector and Option disclosure
Every figure is reported with its sector mix and NEC Option mix. Headline numbers without context are not published.
An aggregate without its mix is misleading. Every published figure is accompanied by its underlying sector composition (rail, energy, defence, water, social infrastructure, etc.) and its NEC Option composition (A / B / C / D / E / F). Where the mix shifts materially issue-on-issue, the change is flagged in the editorial commentary.
04
Definitional discipline
Every term is given a published, citable definition. Apples to apples, every issue.
Compensation event, early warning, claim, settlement, time-to-quotation, time-to-assessment, dispute referral, adjudication, recovery rate. Each carries a published definition that contributors apply when submitting data and that readers can cite when using the benchmark. The definitions live in this methodology document, versioned, with change history.
05
Outcome over opinion
Benchmarks measure outcomes — time, cost, dispute rates — not perceptions. Survey-only metrics are excluded.
Industry surveys produce sentiment data: how respondents feel about NEC, what they believe is reasonable. The Benchmark deliberately excludes such metrics. It measures what actually happened on contributing contracts, time-stamped and clause-anchored.
06
Methodology before data
The methodology paper is published nine months before the first issue. A critique window is built in.
This document is published in full nine months before the first benchmark issue ships. Critique is invited from contributors, peer reviewers, and the academic and industry communities. Material critique is responded to in writing and either accepted (the methodology revises) or declined (with reasons stated). The benchmark's integrity rests on the method being defensible in the open before any data is in the room.
07
Versioned, not amended
Once an issue is published, it is not retroactively edited. Corrections ship as a new versioned issue.
A benchmark that is silently amended after publication is not a benchmark; it is a moving target. Once an issue is published, its figures are frozen. Corrections — for data errors, contributor withdrawal, or definitional change — ship as a new issue with a clear changelog. Subscribers retain access to the prior issue.
08
Independence of operator
The Benchmark is operated by a party with no consultancy upside on the result. Findings cut whichever way they cut.
NECCLAUSE accepts no advertising, no sponsorship, and no vendor funding. The publication has no consultancy practice that monetises the findings. Revenue comes from subscriptions and from the Benchmark itself. Findings that disadvantage subscribers will still be published, in the same form, on the same cadence.
The full paper
What the published methodology will contain.
- 01. Scope and purpose
- 02. Eight rules of the benchmark, with worked examples
- 03. Definitions: CE, EW, claim, settlement, time-to-assessment, dispute referral, adjudication, recovery rate
- 04. Contributor eligibility, consent, and withdrawal
- 05. Data schema and submission protocol
- 06. Anonymisation and the k=10 floor in practice
- 07. Publication cadence, version control, and corrections
- 08. Editorial governance and conflict of interest
- 09. Critique mechanism and response protocol
- 10. Roadmap: v0.1 to v1.0 and beyond
Publication timeline
Methodology in Month 9. Data in Month 12.
Now
Methodology preview live
This page sets out the eight rules, the publication cadence, and the contributor framework. The full paper expands each rule with worked examples, edge cases, and the contributor consent statement.
Month 9
Methodology paper published
The full paper publishes alongside the contributor programme. A six-week critique window opens. Material critique is responded to in writing.
Month 9 to 12
Contributor onboarding
Signed consent statements collected. Contract data submitted to the published schema. First k≥10 cells confirmed in advance of publication.
Month 12
Benchmark v0.1
First issue ships. Free. Subsequent issues quarterly. Paid from issue three.
Editorial governance
Methodology Council.
A standing council of three to five named experts will review material methodology changes and adjudicate critique submitted during the public review window. The council will be appointed from independent NEC practitioners and academic construction-law researchers.
Council members are named on the Methodology Council page once appointed. Appointments target Month 6.