You can't run those queries yourself. Google started requiring JavaScript to search in January 2025, which killed plain scraping outright — the maintainer of the main Python library for it says flatly that "HTTP scraping does not work anymore." Run it from a cloud server and you hit a CAPTCHA in one to three requests. Rotating residential addresses buy you 200 to 500 before the same thing happens. So you rent it from someone who has already solved that, and the whole market below exists for that reason.

Here's the kind of work it does:

• Turn 60,000 company names into 60,000 real websites — the join key that everything downstream depends on.

• Check whether a business still exists, or quietly died.

• See what actually ranks for the query your buyer types.

• Watch what a competitor publishes, in the order Google shows it.

None of that needs AI. It's a lookup. It wants to be cheap, boring, and correct.

Nobody is winning on accuracy

I gave seven of these services 116 companies whose websites I already knew and asked each one for the company by name. They all got it right about two-thirds of the time, and the gaps between them were too small to mean anything. Accuracy is not the decision here. Nothing separates them on it.

Which makes sense once you look at what you're buying. Three of the seven — DataForSEO, OpenWebNinja, Bright Data — just resell Google. They pass your query along and hand back what Google says. There's no mechanism for them to disagree with each other.

The bill is where they differ, and it isn't close. The same 60,000-row job costs $36 with one and $420 with another.

For the ones reselling Google, that settles it: identical product, 12x price range, buy the cheapest. The real question is a different one, and it's about the job rather than the vendor.

When you want Exa or Parallel instead

Google gives you links. Exa and Parallel do things Google's results can't, and it's worth knowing which job is which — because I use both, constantly, and the choice isn't about who's more accurate.

Exa takes a URL as the question. Hand it a company's website and ask for pages like this one. No Google API does that — the old related: operator is long dead. It costs $0.007 a call. Fair warning from my own test: point it at clay.com and most of what comes back is Clay's own subpages, so exclude the domain you started from or you'll get a tour of the site you already had.

Exa's Websets gives you rows instead of links. You describe the thing — "Series B fintech companies in Denver" — plus the columns you want, and it searches, checks each candidate against your criteria, throws out the ones that fail, and fills in the columns. That verification step is the product. A search API hands you ten links and leaves the judging to you. It runs $49 a month for roughly 800 verified rows, about six cents each. The catch is in their own FAQ: it stops when it hits its search ceiling, so treat it as a good list, not a complete one, whatever the marketing says.

Parallel charges by the row, not the column. Their words: one run "can populate many output fields — whether you request 1 field or 20 fields, the cost is the same," and failed runs aren't billed. Twenty columns for the price of one, each with its own citation, from half a cent a row. Nothing else in this market prices that way, and it's the most GTM-shaped thing I've seen in a while.

Parallel's Entity Search builds a starting list in seconds. I asked it for Series B fintechs headquartered in Denver and had ten companies back in 2.6 seconds. Three of them weren't fintech. That's not a bug — their docs say results are "ranked but not individually verified" and that it returns your requested count "even when few entities satisfy the objective." So over-ask and filter. If you need each row actually checked against conditions, that's their FindAll product, which only enriches the candidates that passed.

What neither of them can do, and this is the part that matters:

• Tell you what Google shows. They're searching their own index. Rank, AI Overviews, People Also Ask, the local pack — an own-index engine can't report on a page it doesn't produce. Any question shaped like "what does my buyer see when they search for my prospect" has exactly one source.

• Local businesses. Neither has a Maps product. Our own code disables that path entirely for both of them. If you're chasing dental practices or roofers, you need Google Maps.

• Cheap bulk. Exa is roughly 12x DataForSEO per lookup. For 60,000 name-to-website rows, that's the $36 versus $420 I mentioned. Parallel's cheapest search tier closes most of that gap at a tenth of a cent.

The trade runs the other way too. An AI-native index is built to be read by a model, and for open-ended research — the kind where you want everything on a topic rather than the ten best pages — it's the better instrument. The nine research passes behind this post ran on Exa and Parallel, not Google. I'm not telling you to stop using them. I'm telling you they answer a different question, and it costs 12x to ask it.

What Google did in 2025, and why it reshaped this market

Google spent 2025 making its results harder and more expensive to get, in three moves. Each one moved somebody's price.

January 2025: it started requiring JavaScript. That killed plain-HTTP scraping outright and forced every vendor onto full browser rendering — roughly 30x the bandwidth per query, right where it hurts. Alongside it Google rolled out SearchGuard, which doesn't just gate you at the door: it keeps watching how the session behaves. Search Engine Land pulled it apart in January after the lawsuit filings exposed how it works.

September 13, 2025: it removed the setting that returned 100 results at once. You used to ask for a hundred and get a hundred, one fetch. Now you ask for a hundred and get about ten. No announcement, no explanation.

That second one is the one in your bill. A hundred results costs ten fetches instead of one. Every vendor woke up to a cost roughly ten times higher, and they split into two camps over who eats it.

One camp passes it to you. They bill per page, so your hundred-result pull now costs ten pages. Their prices went up about 10x. Traject Data's own write-up called it "close to 8x more than before." Scrapingdog's staff blog independently landed on "10 times more expensive." That camp is DataForSEO, SerpHouse, Scrapingdog, ScrapingBee, Apify, Zyte, Scrapfly, Decodo.

The other camp eats it. They bill per result or per query, they still make the same ten fetches behind the scenes, and they don't charge you for them. Serper, OpenWebNinja, Nimbleway sit here. Their prices didn't move.

Neither one is cheating. Same physics, opposite bills. That single split is why one vendor quotes you $0.0006 and the next quotes $0.007 for what looks like the same thing.

December 19, 2025: it sued one of the vendors. Google took SerpApi to federal court in California — its own announcement is here, and The Verge, Ars Technica and Search Engine Land all covered it the same day. The claim is worth understanding: it isn't "you broke our terms of service." It's anti-circumvention under the DMCA — the same law that makes breaking a DVD's copy protection illegal. Google is arguing that getting around SearchGuard is itself the offense, which sidesteps the messy question of whether search results can be copyrighted at all.

SerpApi moved to dismiss in February, arguing search results are facts and facts can't be owned. Arguments were heard June 30. As I write this, the docket shows no decision, and one is expected within weeks.

Here's the part that makes the whole market strange. Google's own official search API costs $5 per 1,000 queries — up to 8x what the scrapers charge. It caps you at 10,000 a day. It's closed to new signups. It dies on January 1, 2027. And Google's own guidance tells you to "explore third-party providers" — the category it is currently suing.

So the split is clean. If the job is "understand this market, build me a list, enrich these rows with cited columns," reach for Exa or Parallel — that's what they're built for, and it's why every research pass behind this post ran on them.

If the job is "what does Google say," you're buying a commodity, and the only thing left to decide is who you buy it from.

For those who are paying me COLD HARD CASH, here are the winners…

Read more

Last month I pointed a 63-agent research sweep at every category of data I could think of — regulator rosters, bulk filing systems, city open-data portals, company registries, job feeds, web-crawl archives, cheap paid APIs. 307 sources survived link-checking. 22 died on contact. The census layer — which registries hand you a complete company list — became 316 Free Datasets Build Every List I Sell.

This post is everything else that made the cut: the sources that hand you a named person, a dollar amount, a dated event, or a live feed — free, or close enough that the difference stops mattering. Three of them deserve the full story. The rest of the shelf is named below, category by category, and the link pack — every verified URL, working endpoint, join key, and per-record price — sits behind the line.

1. The landlord behind the LLC

New York City forces the LLC shell open. Every residential building with three or more units has to register with the city's housing department, and the registration must name human beings — corporate officers, agents, site managers — each with a business address. The city publishes the whole thing as an open-data file, one row per person, joined to the building. Rows were last updated June 1.

"Who owns this building" usually dead-ends at "124 Main Street Holdings LLC." Here, the law made someone sign. If you sell to landlords or property managers — proptech, insurance, security, contractors — that's a named-human TAM for the price of a download.

The play: filter the contacts file to corporate officers, join it to the registrations file, and cross the building's tax-lot ID into the rest of the city's property stack — violations, permits, liens. A pain-qualified landlord list with a person's name on every row.

Chicago runs the same play for businesses: its license data ships with a sibling file listing the human owners behind every licensed account.

2. Your competitor's customer list, refreshed monthly

Every US employer sponsoring a retirement or benefits plan files a Form 5500 with the Department of Labor, and the bulk files are free. The sponsor record alone is a private-company TAM: name, EIN, address, phone, and participant counts that work as a headcount proxy for companies no other filing touches.

One schedule inside — Schedule C — names the service providers each plan actually pays: the 401(k) recordkeeper, the benefits broker, the insurer, with compensation attached. If you sell against any of those vendors, the government publishes their customer list and refreshes the current-year file monthly.

The play: pull the current-year zip, filter Schedule C to a competitor's name, and you're holding every plan sponsor that pays them — with the sponsor's phone number and a headcount proxy in the same filing. Displacement targeting from one free download.

3. Executives at companies raising money, before it's news

When a private company raises under a Reg D exemption, it files a Form D — and the SEC publishes quarterly structured extracts of every filing: issuer, industry group, revenue range, amount raised. The current extract runs through Q1 2026.

The hidden gem is the related-persons table: officers, directors, and promoters, with addresses, at companies most sellers haven't heard of yet. "Just raised" is the most crowded signal in outbound — every congratulations-on-the-round email proves it. This is the same signal from the primary source: it includes the raises that never got a press cycle, and it names the people instead of the logo.

I've run this movie before — the franchise build turned 1.34 million public filing rows into 142,579 named operators. Filings beat databases because filings are where the truth gets signed.

The rest of the shelf, in one pass

Paychecks and medical money. Sponsor an H-1B or PERM case and the Department of Labor discloses it quarterly — employer name, worksite, job title, wage. CMS publishes every payment from drug and device makers to physicians, PAs, and NPs back to 2013, with the paying company and the amount on each row, and its nursing-home data names the entities and individuals behind every facility with ownership percentages.

Named decision-makers in routine files. The credit-union regulator's quarterly financial zip carries a CEO-name field next to each institution's website and assets. The FAA's aeronautical file lists every US landing facility with the owner's and manager's name, address, and phone. The federal higher-ed universe ships each college's sitting chancellor or president by name.

Licensed-professional rosters. FINRA runs a free, no-key JSON API over every registered broker — a query for "smith" returned 10,625 reps while I drafted this. The SEC posts monthly structured extracts of every registered investment adviser, the IRS publishes every active tax-return preparer twice a year, and the patent office posts every registered patent attorney and agent. I built a TAM of named contacts for $0 on exactly this pattern — for a regulated vertical, the roster IS the list.

Hiring signal without a job-data vendor. A hiring posting is a public admission of pain. Three of the major applicant-tracking systems publish keyless per-company JSON endpoints — my test pull returned 510 open roles from one company's board, and one of the three returns compensation bands if you ask. Common Crawl's free index enumerates the company slugs no directory lists. One GitHub repo publishes roughly 967,000 currently-open postings across 16 systems as a single public-domain file — though it claims daily refresh and hadn't updated since June 25 when I checked, so archive what you pull. For calibration: Greenhouse reports about 7,500 customers (a "100,000 companies" figure circulating in one research report is flat wrong), and the enterprise systems that dominate large employers — Workday alone holds 38.6% share — publish no public endpoints at all. The inverse signal is free too: an open-source scraper collects state WARN layoff notices on demand, with a ~42,000-notice pre-collected snapshot for backfill. And the Indeed publisher API in a hundred tutorials is dead.

The web's own records. A public archive on Google's data warehouse records millions of sites' detected technology stacks — "every site running a given ecommerce platform" is one SQL query inside the free monthly scanning tier. The internet's naming authority hands out free zone-file access covering 1,100+ domain endings; day-over-day diffs are a birth-and-death feed for company websites (budget weeks for approvals). Certificate logs date a site's launch and leak unannounced subdomains, and WHOIS has a structured, keyless JSON replacement.

Feeds and abroad. An open-source project already scraped every brand's store locator — 40.9 million exact branded locations in the July 4 run. A public project rebuilds the full Y Combinator portfolio daily with an active/acquired/dead flag. A research index answers keyword queries over global news every 15 minutes, free, no key. Apple's keyless app-store API turns "every company with an iOS app of type X" into a TAM, where a stale update date flags a neglected product. Insurance carriers file rate and product changes into a public multi-state system before the market reacts. The UK regenerates a daily file of every company's beneficial owners, and US customs sea-shipment records are searchable free, importer by importer.

The fraction-of-a-cent tier. Where free ends, the paid tier I actually run starts at prices most teams don't believe: Google Maps businesses at $4 per 1,000 on a free scraping plan (sliding to $2 and below on paid tiers), a million email addresses verified for $449, Google search results at six hundredths of a cent per page, any URL turned into clean LLM-ready markdown with ten million free tokens on a new key, live WHOIS at $2 per 1,000, and structured cited answers at $0.006 a call. The link pack carries the exact tools and prices — including one actor that vanished from the store between my July 3 link-check and today, which is its own lesson.

Where free actually ends

Four independent research streams — my swarm plus three external deep-research runs — converged on the same boundary. Free public data reliably delivers the entity, its domain, its address, its legal identifiers, its licenses and filings, and often a named owner or executive. It does not deliver verified work emails at scale, direct-dial mobiles, buying intent, or complete org charts. Spend money there and only there, on rows the free gates already passed.

One legal flag to keep: FEC contribution records carry each donor's self-reported employer and occupation — the most tempting people-to-company map in public data — and federal law bars using them for solicitation. Validation only. Some free data is free with teeth.

— Written by Claude Fable 5, Approved by Jordan

Who Gets This

Everything above is real and checkable — the tour is free. Below the line is the part that saves you the hunt.

• Free: the three flagship files in full, the whole shelf named, and where the boundary of free sits.

• $50/mo (most readers start here): the link pack — every source above with its verified URL, the working endpoints, the join keys, the per-record prices, and the gotchas that cost me real time.

• $2,499/yr: Every tool I ship. Edge Copilot is how you talk to all of it through Claude Code. Current tools: Edge Copilot, AutoClaygent, Agent 7, Who to Target and What to Say, Blueprint Cloud, Technology Finder, Video List Extractor, Competitor Monitor, LinkedIn Engagement, Domain & LinkedIn Finder, Dossier Builder, PDF Contact Finder, TAM Contact Harvester, Find a Rep, Blueprint Playbook, and Crawford. Whatever ships next is included. Plus all 3 courses + weekly office hours.

Below the line, every source above becomes a link pack: the verified URL for each one, the copy-paste endpoints (FINRA's live JSON API, the three ATS job feeds with test companies that actually return data), the join keys that stitch files together (registration ID to tax lot, EIN plus plan number), live per-record prices on the seven paid tools, and the gotchas that cost me real time — which government sites block scripts and where to pull the files instead, and which "refreshed daily" claims aren't.

Start at $50/mo — get the link pack

Go annual — $2,499/yr, every tool I ship

Read more

Every prospect list you've ever bought was a photocopy of a photocopy. Someone scraped a source, resold it, and by the time it reached you the names were stale and the lineage was gone. The original is free. It sits in government file rooms, it has your buyers in it, and most of it carries a number that lets you join one file to the next with zero guessing.

I built an agent named Crawford to work that original. It knows 316 public data sources cold — where each one lives, what it costs, what it joins to, and which ones are dead ends that waste a call. This post ranks the best of them, from the mildly useful to the ones that make a whole market fall out in an afternoon. The sort order is the point: the further down you read, the better the data gets.

Here's the rule Crawford ranks by, so you can rank your own.

What makes a dataset worth money

A dataset's GTM value comes down to five things, multiplied:

• Ground-truth distance. How close is this to the actual event? A government registry that recorded a real license beats a database that guessed at it.

• Join-key quality. Does every row carry a unique ID — an NPI, an EIN, a license number — that lets you connect it to other files without fuzzy name-matching?

• Coverage. Is it the whole universe, or a self-selected slice?

• Free bulk access. Can you download the entire thing, or only peek one record at a time?

• Update cadence. Refreshed weekly, or frozen in 2021?

The first factor is the one nobody internalizes. Public data sorts into a ladder by how far it sits from the real world:

1. Government registry with a unique ID — ATF firearms licenses, the NPI provider file, TTB alcohol permits, a state license roster. Someone had to legally register, and got a number.

2. Legally-mandated disclosure — SEC filings, franchise disclosure documents, state business registrations.

3. Certification and professional directories — board certifications, trade-association credential lists.

4. Platform data with real-world grounding — Google Maps, OpenStreetMap. Proves a storefront physically exists.

5. Association membership lists — self-selected, usually 10 to 40% of the real market.

6. Scraped aggregators and B2B databases — resold, stale, and you can't see where the data came from.

Most vendors sell you rung six and call it a database. The census you actually want lives on rungs one and two, and it's public.

The join key is the other thing that decides everything. A clean unique key outranks any richness a source is missing. A healthcare list I built once joined three separate federal files on the NPI number with zero fuzzy matching — provider, prescribing volume, and pharma payments, stitched together in one pass. A different build with no shared key needed heavy entity-resolution to guess which "Smith Plumbing" was which, and still produced a 196,000-row over-merged mess I had to split back apart.

Keep the ladder and the key in your head. They explain why the free stuff below is fun but limited — and why the paid stuff is where lists actually get built.

The honorable mentions

These datasets are real, they're free, and I use them. They also sit low on the ladder, so I'll be quick. Watch the numbers climb as we go.

The lead dumps

There's a site called emtoss that gives away 452 category CSVs — more than 59 million business leads, free. The catch is the email column: that's the paid add-on. People Data Labs publishes a free company file with over 22 million companies. Both are big, both are keyless, and both are rung six — resold firmographics with no way to prove where a given row came from. Fine for a sample. Useless as a census.

Update, July 12: emtoss went dark four days after this post went out — the domain no longer resolves. A lead dump you don't control can vanish overnight; rung six in one sentence. The People Data Labs file is still live.

Yelp's Open Dataset is more honest about what it is: 150,346 businesses and 6,990,280 reviews, released for research. Real data, real reviews, tiny slice of the map.

The map dumps

Overture Maps publishes a places file that a lot of people quote at "75 million-plus" locations. Overture's own June 2026 release notes put it around 68.9 million, and the file is a blend of four upstream sources — Meta's 58.9 million, Microsoft's 7.4 million, Foursquare's 6.6 million, and the open AllThePlaces scrape. It carries a stable ID called GERS. It proves a place exists. It won't tell you who owns it or what they buy.

That's the ceiling on rung four: platform data confirms a storefront is real, and stops there.

The join spines

Two free sources are worth knowing not as lists but as glue. Wikidata holds 882,874 items in its business class — I counted them live on July 3rd — and many rows carry a company's website, its LEI, its EIN, and its SEC filer number all at once. GLEIF publishes about 3 million legal entities worldwide, each with an LEI. Neither is a market. Both are how you bridge one market to another when the IDs don't line up.

The dead ends I keep so you don't chase them

Half of Crawford's value is knowing what's broken. A few that look free and aren't:

• HIFLD Open was the go-to free hub for U.S. facility rosters. The public portal got discontinued. The links people still pass around are dead.

• The DEA registrant file gets cited constantly as a free way to find prescribers. It is not free.

• Google Places looks like a TAM backbone until you read the terms — you're not allowed to store the names and addresses you pull.

Naming the dead ends is the cheapest credibility a data person has. Now the gradient tips over.

— Written by Claude Opus 4.8, Approved by Jordan

Who Gets This

Everything above, anyone can find with an afternoon and a search bar. Below the line is the part that actually builds a market — the datasets with a real primary key, sorted good to devastating, ending on the federal registries I reach for first.

• Free: the ranking rule, the honorable mentions, and the dead ends worth skipping.

• $50/mo (most readers start here): the heavy hitters — building footprints, state business filings, parcel rolls, and the clean-key federal registries. What each joins to, the gotcha nobody warns you about, and one play it enables.

• $2,499/yr: Every tool I ship. Edge Copilot is how you talk to all of it through Claude Code. Current tools: Edge Copilot, AutoClaygent, Agent 7, Who to Target and What to Say, Blueprint Cloud, Technology Finder, Video List Extractor, Competitor Monitor, LinkedIn Engagement, Domain & LinkedIn Finder, Dossier Builder, PDF Contact Finder, TAM Contact Harvester, Find a Rep, Blueprint Playbook, and Crawford. Whatever ships next is included. Plus all 3 courses + weekly office hours.

Read more

There is no public list of who owns the franchises in America. Ask "who owns the three McDonald's near me," or "find me every multi-unit Wendy's operator in Texas," and you hit a wall. The big data vendors — FRANdata, FranChimp — sell a version of that list. Everyone else does without.

But the raw material is public, and it's free. Per the FTC's Franchise Rule, every franchise brand has to file a disclosure document with state regulators before it can sell a franchise. Buried in that document, Item 20, is a roster: the brand's franchisees, by name, city, and phone. One filing covers one brand's entire national footprint. The list nobody sells is sitting in paperwork nobody reads.

So I read it. All of it. The result is 142,579 multi-unit franchise operators — the people who own more than one location — built from public filings for about zero dollars. Here's how, and exactly what's in it.

What is a multi-unit franchisee, and how many are there?

A multi-unit franchisee is one operator that owns several franchised locations — sometimes five, sometimes five hundred. They're the real buyers in franchising: the operators with capital, staff, and the appetite to add units. I could name 142,579 of them in the US from public filings alone.

Counting them is harder than it sounds, because one operator hides behind many companies. A single Burger King franchisee files each restaurant under its own LLC, and the same restaurant gets re-disclosed in every state the brand registers in. The largest real operator in the data, Boddie-Noell, runs about 636 units. Carrols runs 575. Resolving that sprawl into real operators is the moat. The data is free; assembling it is the work.

Get the merge wrong and you manufacture monsters. An earlier pass over-merged phone-bridged rows into one phantom "operator" with more than 400,000 units — a magnet that swallowed the real whales. Dissolving it brought the top of the list back to actual franchisees, none over 750 units. Shared registered agents that sign state paperwork for hundreds of unrelated brands get demoted, not counted. That cleanup is the difference between a clean list and garbage.

Are franchise disclosure documents public?

Yes. A Franchise Disclosure Document — the FDD — is a legally required filing. Before a brand can sell you a franchise, it has to register the FDD with state regulators and hand it to prospective buyers. Roughly 14 states run their own registration portals; NASAA's national depository covers the rest. The documents are public records.

Item 20 is the part that matters here. It lists the brand's current and former franchisees, with contact information. Every franchisee has a phone number on file. The franchisees live in all 50 states, but the paperwork sits in a handful of portals — so coverage is a function of which brands have filed, not which geography you're hunting. Scrape a few free sources and you get operators nationwide.

Can you find out who owns a franchise?

Yes — you read the brand's FDD. One Item 20 is that brand's whole national roster — every operator, in one document. That's the asymmetry: the question feels like a 50-state manhunt, and the answer is a few thousand documents that already exist.

This is what the long tail of franchise searches is really asking. "Who owns the Taco Bell on Main Street." "List of Wendy's franchise owners." At scale, a dataset of 142,579 operators with a phone on nearly every one answers that question by the hundred-thousand. The phone is the universal contact — it's on ~100% of records, because the filing requires it. Owner names are thinner, and I'll show exactly how thin below.

Why is most franchise data wrong?

Because the lists everyone sells are frozen. The largest free archive of FDD text is a 2023 snapshot. Buy a franchisee list from most vendors and you're buying a moment that's two or three years stale — operators who sold, brands that folded, phones that changed.

I pulled current filings instead. The national depository and a free public FDD library both publish 2024–2026 documents, and feeding those through the same extractor stamps each operator by the vintage of the filing it came from. 65% of the 142,579 operators show up in a filing from the current cycle, not just the frozen 2023 archive. That freshness is the whole edge: a live operator with a working phone beats a 2023 record every time.

Even my own first cut had a freshness bug, and I'd rather tell you how I caught it than pretend it didn't happen. The extraction mislabeled half the roster as former franchisees — a stray header near the top of a filing was tainting every row beneath it. I ran eleven Claude agents over the source PDFs, page by page, checking the extraction against what each filing actually said. They caught it. The fix cut the mislabeled-former rate from about half to under six percent. Check data like this against the real documents, or you'll ship a number you can't defend.

Then I had the finished list audited end to end by an independent pass — roughly 5,800 judged samples, every Texas registry citation re-fetched, every defect re-verified before any fix. It collapsed two phantom duplicate operators, repaired owner names that had been filed in the wrong order, pulled out a batch of business names masquerading as people, and re-keyed the outlets file so all 421,774 of them tie back to a real operator. The numbers here are what survived that pass.

How do you get FDD data without paying FRANdata or ZoomInfo?

You assemble it from free sources, and you pay a vendor for nothing the public record already gives you. Every expensive step in this build has a free substitute:

• The documents: a public, MIT-licensed Hugging Face archive of 22,986 already-read FDD texts — more than 1.5 million roster rows once deduped — plus current filings pulled free from the national depository and a public FDD library. No document purchase, no scanning bill.

• The contact enrichment: a flat-rate enrichment tool that's unlimited on my plan, run across every operator instead of rationed to the top of the list. Zero marginal cost per lookup.

• The store addresses: the Overture Maps open places dataset — 3.9 million US business locations — joined locally to put a street address and map pin on the outlets. The free stand-in for a paid places vendor.

• The operating company behind the storefront: data from the SBA's published 7(a) and 504 loan records. Every franchise-coded small-business loan names the borrowing company and its address. 103,411 of them, free, naming the operating entity and address where the web shows nothing.

The moat was never the data. It's the assembly — pulling the rosters out and untangling the shells into real operators. You can't buy that, and it's what makes the rest free.

What's actually in the dataset

I'd rather show you the real coverage than sell you a clean story. Some fields are dense. Some are thin, because free public data has a ceiling, and I'd rather you know where it is.

• Operators (own 2+ locations): 142,579

• Phone (from the filing): ~100%

• Confirmed in a current 2026 filing: 65% (93,610)

• Named human owner: 16% (22,469)

• Operating entity named (SBA company): ~1% (1,603)

• Mapped outlets (street + pin): 421,774 outlets, 40% addressed

The phone is universal because the filing requires it. The named human owner is the hard ceiling: most franchisees present to the public as the brand storefront, so a phone reverse-looks-up to "Panda Express," not to the operating company behind it. A named human owner sits on 22,469 of the operators — 16% — and I'd rather show you exactly where each one comes from than inflate the number. Three sources, all checkable: on 18,610 the operating entity is itself a person's name, on 2,395 the filing phone reverse-resolved to a named person, and on 1,464 a Texas Secretary-of-State officer record names them. I tried a fourth — matching owners off LinkedIn — and an independent audit found it wrong about half the time, so I cut the whole class. The owners that survived measure at least 95.7% accurate. Naming the rest at scale is a slower, paid job — a Secretary-of-State officer sweep across more states — and it's the next build, not this one.

The operator list is the gold. The phone makes it usable today. The named owners are the bonus where the public record happened to expose them.

Skip the build — download the finished dataset

Everything above is packaged and ready. Annual members pull the full dataset into Claude Code with one command:

• 142,579 multi-unit operators — the headline list, whale-ranked, a phone on ~100% of records

• 421,774 mapped outlets — every owned location, street address + map pin where available

• 22,469 named human owners — each tagged with the public source that named them

• 1,524,943 raw roster rows — the full Item-20 universe the list was built from

Run /edge data franchisee-finder once your license key arrives — it downloads straight into Claude Code and feeds the tools that act on it (--full adds the raw outlet and roster files).

Buy now — go annual, $2,499/yr

Below is the geeky version. Copy it into Claude Code and rebuild the whole thing yourself.

Or don't. Annual subscribers install the tool I actually built with one command — every tool I ship, all 3 courses, weekly office hours.

→ Go annual — $2,499/yr · Start at $50/mo (most readers start here)

Read more