How to Use SEC Full-Text Search to Find Competitive Intelligence

Every US public company has to tell the SEC about its risks, lawsuits, major customers, and strategy shifts, and misstating any of it carries legal liability. All of that disclosure lands in EDGAR, which is free. The standard way to use EDGAR is to look up one company and read its latest 10-K. That works, but it treats the system like a filing cabinet.

The tool worth learning is EDGAR full-text search. It runs one query across every document filed since 2001, from every registrant, all at once. Instead of asking what a single company filed, you ask which companies, out of the entire public market, are talking about a specific thing. For competitive intelligence, the second question does most of the heavy lifting.

A decent chunk of what paid intelligence products sell is this same data with better packaging. Here's how to work the raw source directly.

What the full-text search actually covers

The index covers filings from 2001 to the present across every registrant: 10-K annual reports, 10-Q quarterlies, 8-K current reports, proxy statements, registration statements, and their exhibits and attachments. The exhibits are where a lot of the good material hides. Material contracts, merger agreements, press releases stapled to 8-Ks, and supply agreements all get filed as exhibits, often with the sensitive numbers redacted but the counterparties named.

One boundary to know up front. EDGAR holds electronic filings going back to the early 1990s, but the full-text index only reaches back to 2001. Older documents are still in the system, you just have to find them by company and read them directly.

Who names your competitor as a risk

Risk factors (Item 1A of a 10-K) go through heavy legal review, so nothing lands there casually. When a company names a specific competitor in that section, legal and management have agreed in writing that the competitor is a material threat. Search a competitor's name, filter to 10-K and 10-Q filings, and you get a map of every public company that considers them dangerous, drawn by those companies themselves.

Run the same search on your own company. If firms outside your assumed competitive set have started naming you, the market is defining your category more broadly than your strategy deck does. If a company you consider a rival never mentions you anywhere, that's also information.

To make it concrete, say you operate in payments and want to understand how the market sees a fast-growing processor. Search the name, sort by date, and read the sentence around each hit. The same processor might be framed as a pricing threat by one filer, a partner by another, and a customer-concentration risk by a third, and each framing tells you something about where the money flows.

Proxy statements add another angle. Companies that benchmark executive pay disclose the compensation peer group they measure against, which functions as the board's self-declared list of comparable firms. It's often more honest than anything in the marketing, since compensation consultants pick those names, and you'll sometimes find a company benchmarking itself against firms it publicly claims to have nothing in common with.

Customers, suppliers, and the private company angle

Companies have to disclose when a single customer accounts for 10 percent or more of consolidated revenue, and many name that customer outright. Supplier dependencies show up in risk factors and the business section for the same reason. So a search for one company's name across all filings shows you its commercial relationships from the other side, disclosed by counterparties who had no promotional reason to mention it.

Inside the documents, three places carry most of this: the business section (Item 1) for named partners and channel relationships, the risk factors for dependency language, and the exhibit list for material contracts. If a supply or licensing agreement was important enough to file as an exhibit, the relationship behind it is worth understanding.

This gets especially useful with private companies, which disclose almost nothing themselves. Their public customers and partners still have to file. Say a private logistics software vendor shows up as a critical system provider in the 10-Ks of three public retailers. From those filings alone you can sketch its anchor customer base, make a rough guess at revenue concentration, and see what happens to it if one retailer walks. None of that came from the vendor; it came from its customers' disclosure obligations.

Watching a theme spread through an industry

One company discussing a topic in its filings is noise. The same topic spreading across an industry's filings over a couple of years means dozens of legal departments independently decided it was material enough to tell investors about, which is a much higher bar than a trend piece in the trade press.

The method is simple: run the same query with the date filter set to one year at a time and note the result counts. It's crude, and boilerplate inflates the numbers, but the shape of the curve is informative. Chart mentions of artificial intelligence in annual reports by year and you can watch the line bend after ChatGPT launched in late 2022. Cybersecurity made a similar journey earlier, migrating from a paragraph in the IT section to a near-universal risk factor.

What you're looking for is the shape, with clearly made-up numbers: say a phrase appears in a few dozen filings one year and several hundred three years later, with the new filers clustered in one industry. Filing deadlines force companies to put material themes in writing on a schedule, so this kind of shift often shows up in EDGAR before the paid market research catches up to it.

The same trick maps regulatory exposure. When a new rule is proposed, search its name. The companies discussing it are telling you they expect to be affected, and their risk-factor language usually sketches how they plan to respond long before any of it comes up on an earnings call.

Litigation, distress signals, and companies in play

Legal disputes leave a long paper trail in filings. Search a patent number, a case name, or an opposing party's name and you'll surface every filer discussing the dispute, each with its own read on the exposure. For patent litigation this is usually the fastest way to figure out who actually has money at stake.

Some phrases are worth standing searches of their own. Companies that hire bankers to explore a sale tend to announce a review of strategic alternatives in an 8-K. Auditor doubt shows up as going concern language. On the buy side, acquirers often discuss their eventual targets for years before a bid, as competitors or partners, in their own filings. A large company that keeps naming a small one in a growing market proves nothing on its own, but it's a legitimate input to a watchlist.

Getting precise with filters and syntax

A few mechanics separate a usable result set from thousands of junk hits:

• Exact phrases. Put quotation marks around multi-word terms. Searching "supply chain disruption" as a phrase behaves very differently from three loose words scattered across a document.
• Form type filters. Restrict to 10-K and 10-Q when you want considered, lawyer-reviewed language. Restrict to 8-K when you want news as it breaks.
• Date ranges. Needed for the year-by-year trend counting above, and handy for limiting a query to the current reporting season.
• Company filters. You can restrict a search to a single filer by name, ticker, or CIK, or filter by the location of a company's principal executive offices.
• Name variants. Companies appear under legal names, brand names, and former names. Run every variant, and be careful with companies whose names are ordinary English words, because the false positive rate gets ugly.

Each search produces a stable URL, so saved queries are easy to bookmark and share with a team.

A monitoring workflow you can actually keep up

Here's the version I recommend, in order:

1. Build the term list. Competitor names with variants, your own company, key technologies and standards, product names, named regulations, and a few distress phrases. Ten to twenty terms covers most businesses.
2. Run a baseline pass. For each term, note the hit count, which companies appear, and the context of the mentions. This takes a few hours once, and it's what makes changes visible later.
3. Automate the re-checking. The full-text search has a JSON API behind the web interface, and EDGAR publishes feeds of new filings. A short Python script that reruns your queries daily and diffs against yesterday turns this from a research project into a feed. The SEC asks automated clients to identify themselves with a contact address in the User-Agent header and to keep request rates modest, so build that in from the start. We've automated this at FirmAdapt, but a homegrown script gets you most of the way there.
4. Focus on the changes. A competitor appearing in three new risk-factor sections this quarter tells you more than the forty filings that already mentioned them. New names, new phrasing, and quiet disappearances are the signals worth acting on.

Caveats worth knowing

• The index starts in 2001. Older filings exist in EDGAR but have to be pulled by company.
• Risk factors carry a lot of boilerplate, and copy-paste language inflates counts. Read the surrounding context on a sample of hits before you trust a trend.
• Companies sometimes describe competitors without naming them, especially the ones they fear most. "A large diversified technology company" might be the most important sentence in the section, and the absence of a name is weak evidence either way.
• Some older exhibits were filed as scanned images and index poorly, so zero hits on an exhibit search is not conclusive.
• Coverage is US public companies plus foreign issuers that file with the SEC. Purely private or foreign-listed competitors will only show up secondhand, through counterparties that file.

Where to start

The appeal of this source is hard to beat on fundamentals. The information is verified under legal liability, it covers every SEC registrant, and it costs nothing beyond your time.

For a concrete first step, pick two competitors and one technology term that matters to your business. Run the three searches, filter to 10-Ks from the last two years, and read the ten most relevant hits along with their surrounding paragraphs. Bookmark the search URLs when you're done. That's about an hour of work, and it will tell you whether this deserves a permanent spot in your research routine.