The state of US financial provider evidence
A July 2026 catalogue audit measures what ServeAssess had verified, what remained incomplete and why missing fields describe the research backlog rather than the market.
Identity was much easier to verify than provider detail.
The two bars use different, published thresholds and should not be read as provider quality scores.
Every percentage and rate belongs to the sample, threshold and period shown beside it. A descriptive relationship is not presented as causation.
A verified identity is the beginning of the record
The audit found that 75.6 percent of records reached a verification score of at least 75. Only 3.2 percent reached a research completeness score of at least 40 under the model used for the snapshot.
Those measures answer different questions. A regulator record can establish that an entity exists or reports financial data, while price, customer eligibility and operational outcomes may remain unknown.
Category counts overlap
The catalogue contained 38,010 category assignments across 13,574 providers, an average of 2.8 assignments per record. Category totals cannot be added to estimate unique providers because one legal entity may offer several services.
The overlap also creates an entity resolution problem. A brand may use a bank partner for one product and a separate lender or administrator for another.
Website matching improved usable detail
The audited export matched an official website for 65.0 percent of records. Records with a matched site tended to contain more customer facing detail, although the association does not prove that a provider was more transparent or performed better.
Provider pages remain lower trust evidence for claims that require regulator or government confirmation. They are still useful for current product terms, eligibility and access channels when the observation date is retained.
Missing geography described our backlog
The snapshot placed 47.5 percent of records in a headquarters state repair backlog. That figure measured unresolved catalogue fields and should not be interpreted as a claim about where United States providers are located.
A state page is therefore a research lens. It does not guarantee product availability or permission for a resident or business in that state.
What changed after the audit
Rating Model v3 separated verification strength from research completeness and added entity roles, product permissions and peer group outcome dimensions. Missing outcome data now produces an insufficient data result instead of borrowing meaning from source volume.
The snapshot remains a dated record of the earlier catalogue. Later provider and category totals should not be substituted into its historical denominators.
How the audit counted evidence
The export counted public provider records once by organization and retained overlapping category assignments separately. Linked sources were counted at the profile level, so one official record could support several claim level facts without becoming several independent sources.
Verification and completeness bands were descriptive thresholds from the assessment model used on the snapshot date. They were not estimates of market quality, customer satisfaction or regulatory risk.
Reproducibility and later updates
The report linked its public export and stated the observation date so the aggregate figures could be reconstructed. A later export can differ because records are added, merged, archived or enriched after the snapshot.
Future audits should publish a new dated edition rather than overwrite this denominator. That keeps improvements in collection visible without changing the historical result.
Open the underlying records
Every result on this page belongs to the dated sample above. Later catalogue updates do not silently change its denominator.