The modern information economy produces research reports the way rivers produce water. Every bank, consultancy, think tank, NGO, industry association and now every large company publishes findings, projections and frameworks, usually as polished PDFs with confident covers. A small number are excellent. Many are competent within limits. A meaningful share is marketing wearing the vocabulary of inquiry.
The reader's task is not to reject the category but to grade it quickly. The ten minute examination below, applied in order, does most of the work.
Minute one: the money and the mandate
Look for three disclosures, usually in small type near the front: who funded the work, who wrote it, and what the publisher's mission is. Then ask a question the document will not answer for you: what does this institution need to be true?
That question is not cynical. A trade association publishing on its own industry, a consultancy selling services in the market it analyses, a government aligned institute examining its patron's policies, each has a stake that shapes selection long before a number is fabricated. Disclosure does not disqualify; it calibrates. Undisclosed stakes found later, in contrast, are the single strongest predictor of a report being quietly unreliable.
Minutes two and three: the methodology, read as a confession
Every report has a method, whether or not it admits one. The section, or its absence, tells you which kind you are holding.
A serious methodology names its data sources specifically, with dates. It states the period covered and acknowledges where data ends. It defines its population, which companies, which countries, which years, and keeps that population consistent through the whole document. It shows its arithmetic where it derived new figures, at least enough to recheck one or two.
A weak methodology speaks in genres: "extensive data", "proprietary modelling", "expert analysis", "a comprehensive review". Those phrases are not lies; they are covers for the absence of checkable structure. The test that never fails: could a competent stranger, given this section, reproduce the headline finding? If the answer is no, the report is asking to be trusted rather than verified, and trust is precisely what it has not yet earned.
Minute four: the headline against the evidence
Find the strongest claim in the summary, then find where in the body that claim is actually demonstrated. This sounds trivial and fails remarkably often. The summary says demand will triple; the body shows a survey of nine executives. The summary says the policy destroyed jobs; the body shows a correlation with a paragraph of hedging the summary dropped.
Watch for the specific upgrade move: evidence that establishes association in one section becoming causation in the summary, a finding about one population quietly extending to all populations, a scenario becoming a forecast as it travels toward the cover. The distance a claim travels between its evidence and its headline is a measure of the publisher's discipline, and it is visible in minutes.
Minute five: the counter-evidence test
Ask what would make this report's conclusion wrong, then search the document for whether it asked itself the same question. Good reports contain their strongest objection, because the authors found it first and chose to address it. Reports that contain no serious alternative explanation, no limitation worth the name and no circumstance under which the finding fails, are not confident; they are unrevised.
A related tell: the inconvenient number handled well. Strong work shows the data points that resist its thesis and explains them. Weak work's charts are uniformly supportive, which in real data is close to impossible.
Minute six: the expertise question
Check who actually wrote the analysis. Named authors with published history in the subject, checkable affiliations, and previous reports you can evaluate are the strong pattern. "The Research Team" with no names is the weak one. Neither guarantees anything, but anonymity removes the accountability that lets you hold a publisher to its own record over time, and a publisher's record over time is the best single predictor of its next report's quality.
For quoted experts, one search does the work: what else has this person said, on the record, about this subject? Consistency is not required, evolution is normal, but a source who has argued every side as needed is offering advocacy, not analysis.
Minutes seven and eight: the numbers' paperwork
Units, periods and populations, checked mechanically. Growth against which base, nominal or inflation adjusted, which year's currency, annual against quarterly, global against domestic. Most published errors in this genre are not exotic; they are one of these five mismatches, and reports that commit them usually commit them repeatedly. If you find one, check its siblings before trusting anything else in the document.
Minute nine: what others who read it said
Serious reports get read by serious readers, and the responses are often more informative than the report. A critique from a qualified specialist, a dataset published so others can check, a correction issued when someone found an error: each of these is a good sign about the institution, including the correction. Silence is neutral. But a report promoted only through press releases, quoted only by outlets that received it by email and never by anyone who examined it, is telling you how it expects to be consumed.
Minute ten: decide what kind of document it is
Having done the above, grade the report honestly. It is evidence, and you may cite it as such. It is argument, and you may weigh it against other arguments. Or it is positioning, and you should read it the way you read an advertisement: as a document about the advertiser, not the advertised.
This last grade is not an insult. Institutions produce positioning legitimately. The failure is only in readers treating it as the other thing. The ten minutes above exist to keep that confusion from happening, and they are the same questions asked inside any newsroom that takes sourcing seriously. Reports that survive them deserve to move markets. Most do not.