Reading secretary of state search data with a sceptical eye

The free search screen is the thinnest layer of what a filing office holds. Secretary of state search data begins there and almost never ends there: what you see is an index, a small set of fields extracted from documents so that a name can be found quickly. The documents themselves, the older material, and anything the office keeps but does not publish all sit behind it, reachable by different routes.

Most people never learn this, so they conclude that whatever the search box shows is the whole of the public record. When a question needs more than the index can carry, they either give up or start guessing.

What to take away

Four layers, four ways in

The index. Name, identifier, status, dates, agent, sometimes addresses and officers. Built for lookup. This is what the search box queries and what most commercial copies scrape. In Delaware it is the Division of Corporations' entity search; in California, the Secretary of State's bizfile Online; in Texas, SOSDirect.

Document images. Scans of what was actually filed. Some offices display them alongside the index entry — California's bizfile Online does for many filings — some sell copies, some hold them without publishing them at all. This layer is where the detail lives, because the index stores a summary of a document rather than its contents. Copies come through the office's own records or copies page, where the fee schedule sits beside the order form. Where a document sits in the sequence of filings is set out in filing history.

Certified copies and certificates. Documents the office issues over its own name, dated and identified. A certificate of good standing — also called a certificate of existence — is that same office statement about an entity's current standing, issued for a third party to rely on.

Bulk and programmatic access. Periodic extracts or an interface for machines. Where it exists, it usually carries its own terms, and the licensing questions it raises are the subject of public data APIs. The federal catalog's own guidance on APIs is a fair picture of what a supported interface looks like when a body chooses to publish one.

You will meet all four at some offices and only the first at others. Which of them exists is a property of that office and its budget, not of the law of public records, so check rather than assume.

What the index leaves out

The index is a compression of the file, and compression discards. Expect these to be missing or thin: the contents of documents, since a filing date and a status line say nothing about the terms of an operating agreement; everything filed before the office's imaging begins, which survives on paper or microfilm; and any field the office withholds or redacts, a list that varies by state and often covers personal addresses and dates of birth.

If your question is "what does this company's founding document actually say", the index cannot answer it and no amount of searching will change that. Order the document.

The digitization edge

Every office has a point in its history where the online record thins out and then stops. Entities formed before that point may show a bare index entry with no images, or an entry created when the file was migrated, with an earlier real history sitting on paper or microfilm.

This produces a specific and avoidable mistake: reading a thin online history as evidence that a long-established company has done nothing. It may only be evidence that the office scanned back to a certain year. When a record looks implausibly empty for its formation date, ask the office how far back the imaging goes before drawing a conclusion.

Many offices state the earliest year they have imaged right on the search page, in a note or an FAQ under the results, so the answer is often a scroll away.

What certification actually buys

A plain copy and a certified copy contain the same words. The difference is that the certified one comes with the office's attestation that this is what its records hold, on a stated date, identified so that a third party can check it.

You are not paying for the information. You are paying for provenance you can hand to someone who will not take your screenshot.

Certified copies turn up where certificates of standing do: closings, financings, leases, and registration in a second state. For your own reading, a plain copy or on-screen image is the same document and answers the same question.

Asking for what is not online

Filing offices hold public records, and there is normally a way to ask for one that is not on the website. Every state has a law behind that route — the California Public Records Act, Washington's Public Records Act (RCW 42.56), New York's Freedom of Information Law, Delaware's Freedom of Information Act — and each office publishes the records or public information function that takes the request, usually with a form or a dedicated email address. Finding the right body to address is half the work.

Reaching it through a government path rather than a search result matters here as much as anywhere. USA.gov's index of government agencies is a fast route to national bodies, and each state's own portal does the same job for state ones.

A request lands better when it is specific. Name the entity and its identifier, say which document or which date range you want, and say what form you need it in. "Everything you have on this company" produces delay and often a request that you narrow it. Ask for the certified version only if you actually need certification.

Before you send it, know two things. Offices may withhold or redact particular fields, and what is withheld is a policy decision that varies by jurisdiction, so the absence of a field is not always evidence that nothing was filed.

A request creates a record that you asked, and that record is sometimes the most useful result of the exercise. A documented attempt to obtain something is a stronger note than an undocumented conclusion that it did not exist.

When the copy you hold is not the record

Everything you download, screenshot or export stops being current the moment you save it. That is fine as long as the file says so. Keep the office's own record date next to the date you retrieved it, and keep the identifier, because the entity number is what lets someone else return to the same record later.

The habit that separates a usable file from a folder of images is writing down what you could not get: which document was unavailable, which years are not imaged, which field the office does not publish. Field-by-field, what a record will and will not carry is set out in secretary of state search facts.

Common questions

Why does the office's own copy of a status differ from what an aggregator shows?

Because the aggregator holds a copy taken at some point in the past on a refresh schedule you do not control. OpenCorporates names the official register behind each jurisdiction's data and the date it last collected it; credit and business-data resellers such as Dun & Bradstreet and LexisNexis often do not say. When the two disagree, the office is right about its own register, and the aggregator is right about what the office said earlier.

Can I get a list of every company registered in a state?

Sometimes, through a bulk or data product rather than the search screen, and usually on terms. Several states sell a full extract of their register, and commercial aggregators license bulk data of their own. It is a different request from a records request for one entity's documents, and it is worth being clear which one you are making.

The office shows an index entry but no images. Is the document lost?

Usually not. It is more often unscanned, held in another format, or available only on request. Ask before assuming, and record the answer either way.

Do I need the certified version for an internal credit file?

Rarely. Certification exists so somebody outside your organization will accept the document without trusting you. For your own file, the image plus the record date does the same work.