What is EDGAR full-text search?
EDGAR is the SEC's electronic filing system, and full-text search is the part of it that looks inside the documents. The basic EDGAR company search finds filings by company name, ticker, CIK number or form type. Full-text search finds the words themselves, so you can search for a phrase such as "implementation services" and see every filing that uses it.
The SEC describes it as covering the full text of filings submitted electronically since 2001, including attachments such as exhibits. EDGAR holds older electronic filings too, but those are not full-text searchable, so a search cannot reach anything a company filed before 2001. It is free and needs no account.
How do you search EDGAR full-text?
Start from the SEC's full-text search page, enter a keyword or phrase, and use the date range and filing type filters to narrow the results. Search terms are not case sensitive, but the operators OR, NOT and NEAR must be typed in capitals. The syntax below follows the SEC's search help.
| To do this | Use | Example |
|---|---|---|
| Require every term | Nothing. All terms are required by default. | revenue subscription implementation |
| Match an exact phrase | Double quotes | "standalone selling price" |
| Match either term | Capital OR | "principal versus agent" OR "gross versus net" |
| Exclude a term | Capital NOT or a leading hyphen | "contract asset" -hardware |
| Match word endings | Asterisk at the end of a stem | capitaliz* |
| Match terms near each other | Capital NEAR, with a distance in parentheses to change the default of 10 words | implementation NEAR(5) distinct |
Three restrictions catch people. Writing AND is not needed and is not recognized. The asterisk works only at the end of a word. It does not work at the start or in the middle of a word, inside an exact phrase, or in a search that uses OR, NOT or NEAR. And there is no natural-language search, so a question typed as a sentence returns poor results. Then add a form type and a custom filing-date range to cut the list down.
What filings does EDGAR full-text search cover?
Registrants file registration statements, annual and quarterly reports, current reports, proxy statements and many other forms on EDGAR, and full-text search covers the text of these forms, as filed electronically, from 2001 on. Three kinds of document matter most to accounting teams:
- 10-Ks and 10-Qs. The financial statement footnotes, critical accounting estimates in MD&A, and risk factors. See 10-K vs. 10-Q vs. 8-K for how the forms differ.
- 8-Ks and their exhibits. Earnings releases furnished under Item 2.02, and investor materials furnished under Item 7.01, are usually attached as exhibits (often Exhibit 99.1), so their wording is searchable too. The Regulation FD and Form 8-K guide explains why companies use those items.
- SEC staff comment letters and company responses. These are filed on EDGAR as form types UPLOAD (staff letters) and CORRESP (company responses). The staff releases them no sooner than 20 business days after it completes its review of a periodic or current report, or declares a registration statement effective. Responses are usually filed as CORRESP, but some companies also describe their changes in an amended filing, so a form-type filter alone can miss part of the exchange. You can browse them by topic on Gaapio's SEC comment letters page.
How do technical accounting teams use EDGAR full-text search?
The common uses all ask the same question: what did comparable companies actually do?
- Precedent disclosures. Before filing a footnote on a judgment or a newly adopted standard, find how three to ten comparable registrants presented it, so you can see the range of practice in filed disclosures.
- Comment letter responses. When the staff questions a disclosure, read how other companies answered the same comment, and what the staff did next.
- Auditor and audit committee questions. Show that peers of similar size and industry reached a similar conclusion, as corroboration for an analysis that already stands on the Codification. Filings rarely disclose enough facts to confirm a peer's situation matched yours.
- Adoption of new standards. Transition language has no prior-year template inside the company, so earlier adopters' filings are the closest thing to one.
Peer disclosures show what companies said. They do not show whether a company's facts matched yours, and they are not authoritative guidance. The conclusion still has to stand on the Codification and your own facts.
What does an EDGAR search look like on a real ASC 606 judgment?
Take an illustrative question: a SaaS company sells a subscription with a paid implementation project, and wants to see how peers concluded whether implementation is distinct from the subscription (ASC 606-10-25-17 on set-up activities, and ASC 606-10-25-19 through 25-22 on distinct goods and services).
- Define the judgment in the words filers use. Footnotes tend to say "implementation services", "distinct" and "subscription", and rarely cite paragraph numbers such as ASC 606-10-25-21.
- Run the first query.
"implementation services" distinct subscription, with form type 10-K and a custom filing-date range covering the last two years of annual reports. - Narrow to real peers. The results will include companies in unrelated industries. Keep the SaaS registrants close to your size, which you can identify by their SIC code (7372, prepackaged software, is common, though some SaaS filers use 7370, 7371 or 7374) and the company profile on EDGAR.
- Open each filing at the revenue note. Read the full performance obligation paragraph and the significant judgments disclosure in the revenue note, and check MD&A critical accounting estimates.
- Record what each peer concluded and why. Capture the facts that drove it.
- Check the comment letters. Repeat the search with form types UPLOAD and CORRESP to see whether the staff has questioned any peer's conclusion.
A simple log makes the output reusable in a memo. The rows below are an illustrative format, not findings about any company:
| Peer and period | Implementation treated as | Facts the peer cited | Staff comment? |
|---|---|---|---|
| Peer A (illustrative), FY20XX 10-K | Distinct, recognized as the work is performed | Customers can use the subscription without it, third parties also perform it, and it does not significantly modify the software | Check UPLOAD and CORRESP |
| Peer B (illustrative), FY20XX 10-K | Not distinct, combined with the subscription | Work significantly customizes the software, and the customer cannot benefit from the subscription without it | Check UPLOAD and CORRESP |
| Peer C (illustrative), FY20XX 10-K | Set-up activity, not a promised service | Activities let the company provide the subscription and transfer nothing to the customer | Check UPLOAD and CORRESP |
The point of the log is the middle columns. Two peers can reach opposite conclusions on the same standard because their facts differ, and the memo should say which set of facts yours resembles.
What are the limits of EDGAR full-text search?
- It matches words, not meaning. A peer that describes the same judgment in different words will not appear. You have to guess the synonyms.
- Coverage starts in 2001. Earlier filings are not searchable by their text.
- It returns whole filings to read. Each hit is a document you open and read. A 150-page 10-K may mention a term once, in a paragraph that does not address your question.
- No comparison across companies. Building the peer table in the section above is manual work.
- Hits show only what was filed. A filing the staff never questioned has not been endorsed by the staff.
- Comment letters lag. They appear only after the staff completes its review, so letters from reviews that are still open, or that closed within the last few weeks, are not yet visible.
- Query syntax is strict. No natural language, no leading or mid-word wildcards, and no wildcards inside phrases or boolean searches.
What is EDGAR AI, and how does AI change EDGAR research?
"EDGAR AI" usually means software that applies AI to the filings on EDGAR: you ask a question in plain language, and the tool retrieves and summarizes the relevant passages. It is separate from the SEC's own tools and from what companies disclose about their use of AI in their filings.
| Step | With EDGAR full-text search alone | With AI research |
|---|---|---|
| Framing the query | Guess the filers' wording and operators | Ask the question in plain language |
| Finding passages | Open each filing and find the paragraph | Passages surfaced from the filings, including ones worded differently |
| Comparing peers | Build the table by hand | Draft the comparison for you to check |
| Support | The filing itself | Only as good as the citation back to the filing |
The risk is that a summary reads confidently whether or not the filing says that. Treat an AI answer as a lead to a filing. Open the cited filing, read the paragraph in context, and make sure the peer's facts resemble yours. The accounting conclusion is still yours and your reviewer's.
How does Gaapio help with EDGAR research?
Gaapio is an AI-powered technical accounting platform built by CPAs. For SEC reporting teams, it draws peer comparisons from actual filings, supports comment letter research, and cites its sources, so you can check the filing before relying on it. See how it works on the SEC reporting and disclosure software page, or request a demo and bring a disclosure you are researching now.
Frequently asked questions
See this on your own facts
Bring a real sec reporting question. We'll walk the analysis with you and show you the citation trail.
Related
References: SEC EDGAR full-text search and its search help (sec.gov/edgar/search); SEC Division of Corporation Finance, Filing Review Process (public release of staff comment letters and responses); FASB ASC 606-10-25-17 and 25-19 through 25-22, and 606-10-50-17; SEC, How to Search EDGAR Correspondence; SEC Form 8-K, Items 2.02 and 7.01; SEC Standard Industrial Classification code list. This page is general information, not accounting advice for a specific entity.
Written and maintained by the Gaapio team of CPAs. Last reviewed October 2026.

