Methodology

Most research asks you to trust it. Ours shows its work.

Every claim in a ClearEdge report passes through the same pipeline: it is typed, linked to its source, risk-classified, and reviewed before publication. This page describes that pipeline in full — because a methodology you have to take on faith isn't one.

The claim taxonomy

Seven types. Every single claim in every report carries exactly one of them, and each type has rules about what it's allowed to do.

Claim typeWhat it meansHow reports handle it
Verified factDirectly supported by a primary public source — a filing, pricing page, press release, or record.Published with a numbered citation to that source.
Public statementSomething a company or person said publicly. Saying it doesn't make it true.Always attributed as a statement — never restated as fact.
AnalysisOur interpretation of sourced facts — what a pattern of moves likely means.Labeled as analysis, with the underlying facts cited.
InferenceA conclusion that connects sourced facts but isn't directly stated by any of them.Labeled as inference so you can weigh it yourself.
AssumptionSomething we take as given to make analysis possible.Explicitly flagged — assumptions are never hidden inside conclusions.
AllegationAn unproven claim made by a third party — lawsuits, accusations, rumors.Treated as high-risk: excluded from final reports unless reviewed, approved, and clearly attributed as an allegation.
Low-confidence itemA claim we could not corroborate to our standard.Excluded from final reports by default.

Risk screening

Every claim is risk-classified

Low, medium, or high — based on claim type, source reliability, and subject matter. The classification happens before any human decides what to include.

High-risk claims are gated

Allegations and other high-risk claims are excluded from final reports automatically. Including one requires explicit reviewer approval, and it must be clearly attributed.

Confidence is disclosed

Sources carry reliability ratings (high / medium / low / unverified) and every report carries an overall confidence score. We tell you how sure we are — and aren't.

Source standards

What we use

Lawful, publicly available sources only: company websites and pricing pages, press releases, government and SEC filings, court records, public news coverage, job postings, patents, public reviews, and public statements. Rumor-prone source types — social posts, anonymous reviews — are treated with elevated caution and rated accordingly.

What we never do

  • ✗ No bypassing paywalls, terms of service, robots.txt, or rate limits
  • ✗ No private, restricted, hacked, leaked, or confidential data
  • ✗ No collection of private personal information
  • ✗ No allegations stated as fact — ever

The pipeline

  1. 1. Collect. Public sources are gathered for your scope — competitors, market, questions.
  2. 2. Extract & type. Individual claims are extracted and assigned one of the seven claim types.
  3. 3. Risk-classify. Each claim is scored for risk; high-risk claims are gated automatically.
  4. 4. Human review. An analyst reviews claims and sources before anything is approved for publication.
  5. 5. Generate. Reports are assembled with citations and risk disclosures force-included — the generator will not omit them.

Don't take our word for it

The whole point of this methodology is that you don't have to trust us. Read a full sample report and check the citations yourself.

Read a full sample report