How ProofReports Verification Checks Actually Work

A ProofReports check runs a business's name and any provided details through several independent categories of public records, then summarizes what it found into a single score with a source-level breakdown underneath it. None of the individual checks is a black box - each one maps to a specific, describable question, and each has known limits worth understanding before you rely on the result.

The Shape of a Check

Picture the process as a pipeline with a fixed starting point and five parallel lanes. The starting point is identity resolution: taking whatever you searched - a business name, a URL, sometimes a phone number - and trying to resolve it to a specific legal entity and web presence. Everything downstream depends on getting this step right, which is also where most of the honest limitations live (more on that below).

From that resolved identity, five lanes run in parallel:

1. Registration. Does a matching entity appear in the relevant state's business registry, and what does that record say - active, dissolved, revoked, in good standing? This lane answers a narrow legal-existence question, not a quality question.

2. Licensing. Where the business's trade requires a state or local license - contracting, certain home services, and similar regulated trades - does a license exist for this entity, and what's its current status? Coverage depends entirely on what the issuing board publishes and how cleanly it can be matched to the entity found in step one.

3. Reputation. What do BBB accreditation/rating and aggregated review platforms show for this business, matched by name, address, and website?

4. Web presence. How old is the domain, does the site carry basic trust markers (SSL, a real contact path, consistent business details), and does the online footprint (social profiles, other properties) corroborate the identity found upstream?

5. Government and legal records. Sanctions screening (OFAC), federal court dockets, exclusion status on the federal SAM.gov system, and other public government sources that can surface serious red flags a reputation check alone would miss.

Each lane returns a result and a confidence level, and those roll up into the overall ProofScore rather than being averaged blindly - a strong registration match carries more weight than a thin reputation signal, for instance.

The Honest Limits

Two limits matter more than the rest, and it's worth being direct about both.

Name-matching is the weakest link. Every downstream lane depends on correctly connecting the name you searched to the correct legal entity and the correct license, registration, or review record. Franchise brands, DBAs, and businesses whose storefront name differs from their contract-signing entity are the hardest cases - a real match can come back as unconfirmed simply because the automated matching couldn't bridge the gap between "the sign on the truck" and "the name on the state filing."

Coverage isn't uniform. Not every state publishes licensing data the same way, not every trade is licensed at all in every state, and not every jurisdiction makes court records equally searchable. A missing or unavailable result in any lane means that specific source couldn't be checked - it is not the same thing as a negative finding, and treating it that way is a common misread.

Both of these are reasons to treat a low or partial score as a prompt to look closer, not as an automatic verdict. If a lane comes back unconfirmed, the fastest fix is usually to check the source directly yourself using the exact legal name from the entity's registration record rather than the brand name.

Why Transparency Matters Here

A verification product that hides its methodology is asking you to trust a number without understanding what produced it. The source directory lists exactly what's checked and how each result is classified, and the per-report breakdown shows which specific lane drove the score up or down - so a ProofReports result is meant to be a starting point for your own judgment, not a substitute for it.

What Happens After the Five Lanes Return

Once each lane finishes, the results don't simply get averaged into a final number - they're weighted by how strongly each category correlates with confirmed legitimacy and how reliably it can be checked in the first place. Registration and licensing carry more weight than social-media presence, for instance, because a state filing or an issued license is a harder thing to fabricate than a social profile, and because their absence is a stronger signal than the absence of a Twitter account. A report also distinguishes between a category that returned a clearly negative result (a revoked license) and one that simply couldn't be checked (a state that doesn't publish licensing data in a searchable format) - collapsing those two outcomes into the same score would make the number far less useful, so they're tracked and displayed separately in the underlying breakdown.

Why the Process Runs on Request Rather Than in a Static Database

Public records change, sometimes within days, so a check run once and cached indefinitely would drift out of date in exactly the categories that matter most - license status and registration standing move the fastest of any signal in the pipeline. Running the five lanes fresh on request, rather than serving a stored result, trades a few extra seconds of wait time for a report that reflects what the source actually says right now rather than what it said whenever the record happened to be built.

Frequently Asked Questions

Why did a business I know is real come back with a low score?

The most common cause is a name-matching gap - the entity's legal name in state records differs from the brand name you searched. Try searching the exact legal name from a contract or invoice instead of the storefront name.

Does a missing result mean something bad was found?

No. A missing or unavailable result means that specific source couldn't be checked or matched - often a coverage gap in what a state or board publishes - not that a negative result was returned.

Can the checks be wrong?

Yes, in both directions. Public records themselves can be outdated or contain errors, and matching logic can miss a legitimate entity or, less often, connect the wrong one. Treat any consequential finding as a lead to verify directly against the original source, not a final answer.