Why the diseases are in this order

The five diseases are listed in the order published research says they cost a contractor, most first: rework, the baseline overrun, forfeited entitlements, disputes, non-compliance. Weather rolled into forfeited entitlements: a wet day the contract’s extension-of-time or delay-cost clause doesn’t get claimed is exactly a forfeited entitlement, not a disease of its own. research.demiton.io uses the plain names - cost drift, missed claims, lapsed compliance - for the same three.

This page is the argument for that order. The evidence lives at research.demiton.io: every figure, the paper or report it came from, where in that document to look, how sure we are, and what it isn’t. Anyone can dispute a figure or add a source there, and the same record is on GitHub. Nothing on this page is a figure you should quote; quote the research page, which says where it came from.

Why there is no single number per disease

Which figure applies depends on where the work is. A rework study from Australian civil projects says something about a Queensland job; a Singapore building study doesn’t. So research.demiton.io shows each disease’s spread for the state and industry you pick, and Protection applies to each of your projects the nearest evidence for its own location (the rule). Where no evidence applies, it says so rather than borrowing someone else’s number.

How we set the order

In September 2026 we ran separate literature searches, one per disease. Each had three rules: only report a figure actually read in the source, say exactly what it’s a percentage of, and say so plainly when good evidence doesn’t exist instead of stretching a weak number. Every figure was then checked a second time against the original source. Where we could only see a figure through someone else’s citation, it’s marked second-hand. Those searches are now a public record that keeps growing: how a figure gets in.

Where two diseases had similarly strong evidence, we broke the tie by asking which evidence measures money the contractor itself loses, rather than money that’s merely at stake or mostly borne by someone else. That’s why rework sits above the baseline overrun, even though overruns are bigger in dollars.

The evidence isn't equally strong for all five

Rework has real measurements of what contractors lose. For most of the rest, researchers have measured how much is at stake (how much is claimed, held back or fought over) rather than how much a contractor ends up losing. Every figure on research.demiton.io is labelled contractor loss, money at stake, owner’s money or societal, and only the first two are ever applied to your contracts.

1. Rework

Work done twice: kerbs poured to the wrong level, built off a superseded drawing, redone after a failed test. Rework has the strongest evidence of the five, and it points one way: it’s common, it’s large relative to a civil contractor’s margin, and the contractor pays for it. Surveys of Australian civil projects and one contractor’s own ledgers agree on that much; they disagree on how big, because most rework is never recorded as rework.

Every rework figure, with its source → · How Demiton catches it →

2. Baseline overrun

The gap between the price a job was won at and what it actually costs. There’s strong evidence that overruns are large and common, and that margins are thin enough for a modest one to wipe them out. There’s weak evidence on how much of an overrun the contractor keeps rather than recovering through variations, and much of the best-known overrun research measures the client’s money, not the contractor’s.

Every baseline overrun figure, with its source → · How Demiton catches it →

3. Forfeited entitlements

Variations, extensions of time, delay costs and retention the contract allowed and nobody claimed in time. No study measures the share a contractor never recovers. The evidence shows how big the pot is, how rarely contractors use the payment protections they’re entitled to, and how much is never paid when a business fails. Weather sits here: the evidence is on how many days it takes and what a standing day costs, and on contracts written so the contractor carries most of it.

Every forfeited entitlements figure, with its source → · How Demiton catches it →

4. Disputes

Money fought over, what it costs to fight, and how much is recovered. There’s good evidence on how much is at stake and what the fight costs, mostly from large projects that were already in trouble, and Queensland’s own adjudication record. No verified figure exists for what a dispute costs a contractor as a share of the contract.

Every disputes figure, with its source → · How Demiton catches it →

5. Non-compliance

Accidents, lapsed licences and insurance, and the cost of breaching the obligations that come with the job. There’s solid national data on what non-compliance costs in total, and it consistently shows most of that cost falls on workers and the public, not the employer. Costs per incident are measured; the cost of losing standing altogether (dropped from a licence, a panel or a prequalification list) isn’t. That’s why it ranks last on measured cost, not on how much it matters.

Every non-compliance figure, with its source → · How Demiton catches it →

What this means for you

This order is the default. It decides the order the diseases show up in during setup, and the order we build detection for each one. It isn’t a claim about your business: a contractor with a run of disputed variations, or one heading into a wet season, may reasonably rank them differently. Setup in the app lets you put the five in whatever order matters to you.

If you find a figure that doesn’t match its source, or a better study we missed, suggest it on research.demiton.io.

Next steps

• research.demiton.io - every figure, its source, and what it isn’t
• The five diseases - what each one is, and the registers it reads
• Register catalog - every register, grouped by domain and by disease