Our Investigation Methodology

Project Insider Asia exists to answer one question honestly: why do billion-dollar projects turn out the way they do? Answering it credibly requires a method, and this page describes ours. It is written to be proportionate to what we are — a small, independent editorial platform — rather than to imitate the apparatus of a large newsroom.

How investigations are selected

We select cases against three tests. First, materiality: the outcome must matter — in money, in lives, or in public consequence. We generally look for projects or companies where the stakes ran to hundreds of millions of dollars or more, or where official findings recorded serious harm. Second, documentary depth: there must be a record substantial enough to investigate. Court filings, regulator actions, official inquiries, audit reports and parliamentary records make a case investigable; rumour does not. Third, lesson value: the case must teach something transferable to people who commission, govern or deliver large programmes today. A well-documented failure with no generalisable lesson is history, not an investigation.

We deliberately mix eras and sectors. Aviation, healthcare, banking, government IT and infrastructure fail in structurally similar ways, and the comparisons are part of the point.

How facts are verified

Every material factual claim in a PIA investigation is checked against at least two independent sources where reasonably possible, and against primary documents where they exist. Primary documents — court judgments and filings, accident-investigation reports, regulator findings, National Audit Office and congressional reports, audited accounts — take precedence over press coverage. Where we cite a figure, we prefer the figure from the official record over a journalist's paraphrase of it.

We maintain a working distinction between verified fact and analysis throughout production. A claim is treated as verified when it is established by the documentary record: a conviction, a settled loss figure, an inquiry finding. Interpretation of why events unfolded as they did is analysis, and is labelled as such. Where a fact rests on a single source, or on press reporting without documentary confirmation, we either attribute it explicitly or leave it out.

How uncertainty is handled

Large projects produce uncertain numbers. Cost totals for long-running programmes vary between credible sources because of scope definitions, inflation treatment and what is counted as remediation. Where figures are approximate, we say so — "approximately", "estimated" — and where credible sources conflict materially, we disclose the conflict rather than silently picking the number that fits the narrative. The Sydney Opera House cost and the NHS National Programme for IT total are both reported differently across official documents; our investigations note the range and explain why.

How corrections work

We correct errors, and we do so visibly. Factual corrections are made in the text with a dated note identifying what changed and why; we do not silently edit published investigations. Clarifications that add context without changing a factual claim are noted as such. If you believe something we have published is wrong, write to us with the evidence. The full process, including response commitments, is set out in our corrections policy.

How AI tools are used

We use AI tools as research assistance — for example, to help locate and organise publicly available documents and to draft summaries that a human then checks against the underlying sources — and to generate illustrative imagery. AI tools do not determine our findings, and nothing generated by an AI tool is published without human review against source material. AI-generated images are labelled. The details, including what we do not use these tools for, are in our AI disclosure.

How opinion is distinguished from verified fact

Each investigation separates the established record from our judgement of it. The factual narrative is built from the sources described above and is subject to the verification standards on this page. Sections labelled Insider Take are explicitly analysis: Ramesh Dixit's assessment, informed by more than twenty years of programme and project delivery across Asia, of what the documented facts mean and what practitioners should do differently. The Insider Take is argument, not evidence, and it is written so that a reader who accepts every fact we report may still reasonably disagree with our conclusions.

Conflicts of interest

PIA is editorially independent. We do not accept payment, sponsorship or consulting work from the companies, agencies or individuals we investigate, and we do not take instructions from advertisers on coverage. Ramesh Dixit has worked in programme and project delivery across Asia; where a past professional engagement creates even the appearance of a connection to a subject under investigation, that connection is disclosed within the investigation itself. We have no institutional investors with interests in the sectors we cover. If a conflict arises that cannot be adequately disclosed, we do not publish the investigation.

---

Last reviewed: 1 August 2026 · Author: Ramesh Dixit