Verification Framework
How we classify the confidence level of every data point on this site.
Why This Matters
Expert networks self-report most of their own data — expert counts, compliance practices, AI capabilities, pricing structures. There is no independent auditor for this industry. That means not all claims carry the same weight.
Rather than present everything as equally reliable, we label each data point with a confidence tier so you can judge for yourself what to trust.
The Four Confidence Tiers
Every factual claim in a network profile is assigned one of four labels:
Confirmed through independent, third-party sources — regulatory filings, audited reports, court documents, credible journalism, or official registrations. These facts do not rely on the company's own claims.
Examples: founding year from incorporation records, headquarters from regulatory filings, employee count from LinkedIn or annual reports.
Sourced from the company's own materials — website copy, press releases, marketing collateral, or executive statements. The claim may be accurate, but it originates from the company itself and has not been independently corroborated.
Examples: expert count from company website, compliance policies from marketing materials, service descriptions from product pages.
Derived from indirect evidence, industry context, or pattern analysis rather than a direct source. We make reasonable deductions based on available information but flag these clearly so you know the basis is interpretive.
Examples: pricing model estimated from job postings and industry norms, geographic coverage inferred from office locations and hiring patterns.
Approximate figures or claims that cannot be fully confirmed through publicly available information. This applies to data points where we provide a reasonable estimate based on partial evidence, as well as internal processes or proprietary details that are not disclosed in sufficient detail to verify.
Examples: estimated expert counts from partial data, internal vetting procedures, proprietary AI matching accuracy, approximate employee numbers.
How Labels Are Assigned
Confidence labels are assigned per field, not per profile. A single network profile may have its founding year marked as verified, its expert count as positioned, and its pricing model as inferred — because each data point has a different evidence basis.
We review and update labels when new sources become available. If a previously inferred claim is later confirmed by a third-party report, we upgrade it to verified.
What This Means for You
- Verified data can be cited with confidence in research and due diligence.
- Positioned data is useful but should be understood as the company's own framing.
- Inferred data provides directional insight but warrants additional validation for high-stakes decisions.
- Estimated data is included for completeness but should not be relied upon without further investigation.
Relationship to Sources & Methodology
This framework describes how confident we are in each claim. Our Sources & Methodology page describes where we gather data and how we select which networks to include. The two pages are complementary — methodology explains process, verification explains output quality.
Last updated: March 2026