# Swedish Org.nr Scraper - Organisationsnummer & Contacts (`scrapersdelight/se-orgnr-website-contact-scraper`) Actor Turn Swedish company domains into registry-grade B2B leads from each site's statutory 28 kap. 5 § disclosure: företagsnamn, Luhn-validated organisationsnummer, säte, momsregistreringsnummer, F-skatt, bankgiro, email and phone. No login, no API key. $6 per 1,000 disclosures. - **URL**: https://apify.com/scrapersdelight/se-orgnr-website-contact-scraper.md - **Developed by:** [Scrapers Delight](https://apify.com/scrapersdelight) (community) - **Categories:** Lead generation, Business, Automation - **Stats:** 2 total users, 1 monthly users, 100.0% runs succeeded, 0 bookmarks - **User rating**: No ratings yet ## Pricing $6.00 / 1,000 per swedish company disclosure returneds This Actor is paid per event. You are not charged for the Apify platform usage, but only a fixed price for specific events. Learn more: https://docs.apify.com/actors/running/actors-in-store.md#pay-per-event ## What's an Apify Actor? An Actor is a serverless cloud program that runs on the Apify platform. It has two run modes. In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours, and optionally produces a well-defined JSON output, datasets with results, or files in key-value store. In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server. Apify vocabulary and the platform model are defined once, in the agent quickstart at https://apify.com/agents.md. ## How to integrate an Actor? If asked about integration, you help developers integrate Actors into their projects. You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready. Do not guess an integration path. Every one of them is in the agent quickstart at https://apify.com/agents.md: the Apify MCP server, Agent Skills with the Apify CLI, the JavaScript and Python clients, the REST API, and the account-free path for an agent with no human to sign in. It also carries the rule on stating cost before the first paid run. For examples already wired to this Actor's own input schema, see the [API](#api) section below. Each client library has reference documentation the quickstart does not restate: [JavaScript/TypeScript](https://docs.apify.com/api/client/js/docs.md) (`npm install apify-client`) and [Python](https://docs.apify.com/api/client/python/docs.md) (`pip install apify-client`). # README ## 🇸🇪 Swedish Org.nr Scraper — Organisationsnummer, Säte, Moms & Contacts **Give it a list of Swedish company domains. Get back one registry-grade B2B lead per company, read off the disclosure Swedish law already obliges that company to publish on its own website.** Not a directory scrape. Not a guess. Every identifying field in the output is a statement the business is legally required to make about itself, in public, on its own site — and the `organisationsnummer` is checked against its real Luhn check digit before it is delivered. *** ### ⚖️ Why this data exists, and why it is clean **Aktiebolagslagen (2005:551) 28 kap. 5 §**, quoted verbatim from [riksdagen.se](https://www.riksdagen.se/sv/dokument-och-lagar/dokument/svensk-forfattningssamling/aktiebolagslag-2005551_sfs-2005-551/): > *"Ett aktiebolags brev, fakturor, orderblanketter och webbplatser ska ange bolagets företagsnamn, > den ort där styrelsen har sitt säte samt bolagets organisationsnummer enligt lagen (1974:174) om > identitetsbeteckning för juridiska personer m.fl. Om bolaget har gått i likvidation, ska också > detta anges."* In English: **every Swedish aktiebolag's websites must state its company name, the municipality where the board has its registered seat, and its organisationsnummer — and must say so if the company has gone into liquidation.** Three things follow, and all three shape what this Actor does: 1. **The corpus is the whole Swedish commercial web.** There is no member list to scrape and no index of disclosures. Any Swedish company domain is a candidate. 2. **The `säte` is a statutory field in its own right.** It is the *ort* — the municipality — where the board sits, and it is frequently a different place from the postal address on the contact page. It gets its own column here. 3. **A company in liquidation must say so.** That sentence, published by the company itself, is a B2B signal worth a column too — and an input to filter on. **The duty binds aktiebolag.** An *enskild näringsidkare* (sole trader) is not covered by 28 kap. 5 § and has no organisationsnummer at all — Bolagsverket's own words: *"Enskilda näringsidkare har sitt personnummer som identitetsbeteckning i stället för ett organisationsnummer."* Feed this Actor a mixed Swedish list and a large share of it simply is not in scope. Measured on two independent 400-domain samples: **22.9% of live domains on a high-street sample, 33.9% on a traffic-ranked one** — so the more of your list is `AB`, the closer you sit to the top of that range. Both numbers, and how they were taken, are below. *** ### 🔍 What it does, per domain Plain HTTP (`got-scraping` + `cheerio`) over the Apify proxy, with a real browser only where the measurement said a browser was the *only* route: 1. **Homepage + its footer.** On a Swedish site the 28 kap. 5 § statement sits in the site-wide footer far more often than on a dedicated page. 2. **JSON-LD on the same page** — `taxID` / `vatID` / `legalName` inside `application/ld+json`. Free, exact, and on the seed probe it was the *only* route to the identifier on 2 of 17 reachable sites, both of which render their footer client-side. 3. **Ranked Swedish links.** *Köpvillkor* → *om oss* → *juridisk information* → *allmänna villkor* → *integritetspolicy* → *kontakt*. This is the single biggest lever in the build. The English ladder used by the UK sibling (*terms and conditions*, *legal notice*, *privacy policy*) scores **zero** on every one of those Swedish link texts. 4. **XML sitemap → WordPress REST page index.** 5. Two further rungs that are built, working, and **off by default because they were measured and did not pay**: guessed Swedish paths (`/kopvillkor`, `/om-oss`, `/kontakt` …) and a **real Chromium**. The numbers for both are below — switch them on deliberately, not hopefully. Fields are **merged across pages**: a Swedish site routinely carries the organisationsnummer in the footer, the email on `/kontakt` and the säte on `/kopvillkor`. Turn on **per-field provenance** and every field names the exact URL it came from. **A domain that is dead, walled, publishes no disclosure, or turns out to be a sole trader is never delivered as a lead and never billed.** *** ### 📦 What you get — every field One flat row per company (or grouped objects — your choice, one switch). #### Identity | Field | What it is | |---|---| | `registeredName` | The företagsnamn as the disclosure states it | | `registeredNameSource` | `label` / `suffix` / `prefix` / `is-a-company` — how it was read | | `tradingName` | A varumärke or trading name, when the site names one separately | | `companyName` | Best available name: registered name, else JSON-LD `legalName`, else `og:site_name` | | `legalForm` | Aktiebolag (AB) · Publikt aktiebolag (AB publ) · Handelsbolag (HB) · Kommanditbolag (KB) · Ekonomisk förening · BRF · Ideell förening · Stiftelse · Filial | | `entityKindIndicated` | What the leading digit group of the organisationsnummer **indicates** | #### Registry — the statutory core | Field | What it is | |---|---| | `companyNumber` | The organisationsnummer, canonical ten digits | | `companyNumberFormatted` | The published form, `NNNNNN-NNNN` | | `companyNumberRaw` | Exactly as printed on the page | | `companyNumberSource` | `label` / `label-en` / `legal-form` / `label-loose` / `bare` / `json-ld` | | `companyNumberChecksumValid` | **A real checksum.** The Luhn (mod-10) check digit, not a format guess | | `companyNumberScheme` | `SE-ORGNR` | | `sate` | **The statutory säte** — the ort where the board has its seat | | `jurisdiction` | `Sweden`, or `Sweden (foreign-registered entity)` for a 3-group number | | `registerUrl` | A free public lookup keyed on the same number | | `registeredOffice` / `…Postcode` / `…City` | The postal address stated in the disclosure | | `vatNumber` | The momsregistreringsnummer **as published on the page** | | `vatNumberValid` | Luhn on its ten-digit core | | `vatCore` / `vatBranch` / `vatCountry` / `vatChecksumVariant` | The decomposition | | `derivedVatNumber` | The `SE……01` form **derived** from the organisationsnummer | | `vatMatchesOrgNr` | Does the published momsnummer decompose to the published organisationsnummer? | | `inLiquidation` / `liquidationStatement` | The 28 kap. 5 § second-sentence statement, verbatim | | `fSkattStated` | The site states it is *godkänd för F-skatt* | | `bankgiro` / `bankgiroValid` · `plusgiro` / `plusgiroValid` | Sweden's business payment identifiers, each with its own mod-10 check | **About `derivedVatNumber`.** A Swedish momsregistreringsnummer is `SE` + the ten-digit organisationsnummer + a two-digit sequence. That is verified, not assumed: three real organisationsnummer read off these companies' own pages were put through the European Commission's VIES API on 2026-09-20 and all three came back `isValid: true` with the right company name (`SE556035867201` → *Clas Ohlson Aktiebolag*, `SE556628159701` → *Nordic Nest AB*, `SE202100548901` → *BOLAGSVERKET*). **But having an organisationsnummer does not make a company VAT-registered.** `derivedVatNumber` is the *form* the number takes if the company is registered — it is never presented as proof that it is, which is why it lives in its own column and never in `vatNumber`. **About `entityKindIndicated`.** Bolagsverket publishes the mapping (5 = aktiebolag, filial, bank, försäkringsbolag, europabolag · 9 = handelsbolag/kommanditbolag · 7 or 8 = ekonomisk förening, BRF, näringsdrivande ideell förening · 2 or 8 = trossamfund · 20 = statlig myndighet · 3 = utländskt företag) and publishes this caveat with it, which we repeat rather than hide: *"Det går inte att helt säkert säga vilken företagsform som finns bakom ett organisationsnummer."* The field is named `…Indicated` for that reason. #### Contact | Field | What it is | |---|---| | `email` / `emails` | Deduplicated, on-site addresses first; Cloudflare-obfuscated ones decoded | | `emailConflict` | Set when the best email is on a *different* non-freemail domain | | `phone` / `phones` | E.164 (`+46…`) | | `addressLine` / `postcode` / `city` | The trading/visiting address, when it differs from the registered one | | `socialLinks` | LinkedIn, Facebook, Instagram, X, YouTube, TikTok — profiles only, never share buttons | | `officerName` / `officerRole` | A VD, styrelseordförande, firmatecknare or ansvarig utgivare named next to a label | | `termsUrl` / `privacyPolicyUrl` | The site's own policy pages | #### Provenance & accounting | Field | What it is | |---|---| | `disclosureUrl` | The exact page the disclosure was read from | | `disclosureSource` | `homepage` · `homepage-json-ld` · `legal-page` · `merged` · `browser-render` | | `discoveryChannel` | Which channel found it | | `renderedWithBrowser` | Whether the Chromium rung was used for this domain | | `pagesParsed` / `pagesChecked` | How much work the row cost | | `fieldCount` | How many of the 20 value fields are populated | | `stableId` | The organisationsnummer when valid, else the registrable domain | | `status` | `ok` · `no_disclosure` · `sole_trader` · `blocked` · `unreachable` · `filtered_out` · `duplicate` · `suppressed` · `skipped_robots` · `error` | | `missReason` | Why, in a sentence, for anything that is not `ok` | | `fieldSources` | Per-field provenance (optional) | | `elapsedMs` / `fetchedAt` | Timing | *** ### 🔐 Sole traders and personnummer — the one thing we deliberately do not hand you A personnummer and an organisationsnummer are **both ten digits and both carry the same Luhn check digit**, so a checksum cannot tell them apart. What can: positions 3–4 of a personnummer are a birth month (01–12), and an organisationsnummer is issued with that pair **≥ 20** precisely so the two ranges never collide. When a site publishes a Luhn-valid number whose date pair is below 20, this Actor: - **does not** treat it as an organisationsnummer, - **does not** deliver the row as a company lead, - **does not** charge you for it, - returns `status: "sole_trader"` with a plain-English `missReason` if you asked for miss rows, - and **withholds the digits** unless you explicitly switch on *Include a sole trader's personnummer*, because a personnummer is personal data and that business is not bound by 28 kap. 5 § in the first place. No other Actor on the store draws that line. It is also the honest answer to "why did my Swedish list convert at 35% and not 80%" — a big slice of any Swedish SME list is enskild firma. *** ### 📊 Measured results — real numbers, not claims Every number below comes from a run on the Apify platform through the Apify proxy. None of it is an estimate. #### The test corpus Real Swedish businesses from **OpenStreetMap** — `office` + `shop` + `craft` elements carrying a `website` tag, intersected with the Swedish national area — **4,231 unique hosts**, picked without any reference to whether they publish a disclosure. | TLD | hosts | share | |---|---|---| | `.se` | 3,426 | **81.0%** | | `.com` | 562 | **13.3%** | | `.nu` | 133 | 3.1% | | `.org` / `.net` / `.eu` / other | 110 | 2.6% | **19% of a real Swedish business corpus is not `.se`.** That is why no TLD filter is applied by default. The corpus was then split by a stable hash of the domain into **3,404 development** hosts and **827 held-out** hosts. The held-out slice was never looked at while the parser was being written. The numbers below are from it. #### What 299 held-out Swedish domains returned | outcome | domains | note | |---|---|---| | **billed leads** | **72** | 27.9% of the domains that were reachable | | publishes no disclosure | 174 | never billed | | host did not resolve | 41 | never billed — stale OSM `website` tags | | blocked / challenged | 9 | never billed | | sole trader (personnummer) | 2 | never billed, digits withheld | | duplicate company | 1 | collapsed before billing | **Delivered rows: 72. `chargedEventCounts`: 72. Exactly equal**, with the counters polled to stability — no row was billed that was not delivered, and none delivered that was not billed. The same 299 domains were run three times across three builds and returned **69, 70 and 72** billed rows. That spread is transport luck — dead and soft-blocked hosts move between runs — and it is why the yield below is quoted as a range rather than as one run's number. #### Where the disclosure actually was | found on | leads | |---|---| | merged across two or more pages | 39 | | a dedicated legal page | 31 | | the homepage alone | 2 | | found via | leads | |---|---| | the site's own ranked **Swedish** links | 40 | | homepage footer (free) | 27 | | homepage JSON-LD | 3 | | XML sitemap | 2 | | guessed paths | **0** | **More than half of all leads needed two or more pages merged.** A Swedish site carries the organisationsnummer in the footer, the email on `/kontakt` and the address on `/kopvillkor`. #### Per-field fill, on the billed rows | field | fill | | field | fill | |---|---|---|---|---| | organisationsnummer | **95.8%** | | address line | 86.1% | | derived momsnummer | **95.8%** | | postcode / city | 86.1% | | entity kind (indicated) | 95.8% | | email | **91.7%** | | legal form | 81.9% | | phone | 81.9% | | företagsnamn | 80.6% | | published momsnummer | 18.1% | | registered office | 63.9% | | F-skatt stated | 4.2% | | säte | **4.2%** | | named officer | 4.2% | | bankgiro | 1.4% | | plusgiro | 0% | **Every one of the 69 organisationsnummer delivered passed Luhn.** Zero checksum failures. All 13 published momsnummer passed; 10 of 13 decompose to the same company's organisationsnummer. **About `säte` at 4.2%.** 28 kap. 5 § requires it, and Swedish sites overwhelmingly do not publish it. That is a fact about Swedish compliance, not a parser gap — see the next section, which is how you can tell the difference. #### How you can tell that is not a parser gap Two checks, both run against the raw bytes rather than the output: **1. Label present → value emitted.** Of the 156 captured pages, 52 carry an `org.nr` label in their bytes. **51 of 52 emitted a value. The 52nd carried a personnummer and was withheld on purpose.** The corrected rate is **51/51 — 100%.** Where the field is on the page, it comes out. **2. The negative direction.** Every row with no organisationsnummer was re-swept with an **independently written** Luhn — different code from the one that ships — looking for a labelled, organisation-shaped number the parser had missed. **Zero.** Then the same was done live: 45 domains were drawn at random from the `no_disclosure` bucket and **re-fetched from scratch**. 42 answered. **Exactly one of the 42 carried an org.nr label at all, and zero carried an organisationsnummer the Actor had missed.** The bucket is real. #### We never repair a number to make it check out A Swedish organisationsnummer is unusually easy to fabricate: a random ten-digit run passes the Luhn checksum **by luck about one time in ten**, so "it passes the checksum" is not evidence that it is the number the site published. The rule here is **output ⊆ input** — every digit delivered appears, as those digits, in the bytes the page served. The suite brute-forces it: ``` 4,680 single-digit mutations of the 52 real organisationsnummer in the test corpus 0 were repaired into a different number ``` Alongside it: a bare Luhn-valid run with no label is refused · a labelled number that fails Luhn is refused outright rather than nudged · a momsnummer whose core fails Luhn is published **as found** with `vatNumberValid: false` and never promoted to `companyNumber` · a labelled Luhn-valid personnummer is refused as a company id · and two numbers in adjacent HTML elements can never be welded into one, because the separator inside a digit run is horizontal whitespace only. Normalising `556035-8672` to `5560358672`, dropping the `16` century prefix, or unwrapping `SE……01` to recover the core is **canonicalisation** — no digit is invented, and the test asserts every digit is still in the input. Inferring a missing digit is **completion**, and there is no code path that does it. #### Printed on the page, or declared to machines? A page can carry its organisationsnummer inside a `