Krak Scraper — Danish Business Directory Leads (Virksomheder)

The krak scraper that turns krak.dk and degulesider.dk — Denmark's two Yellow Pages sites — into a clean B2B lead list: company name, phone number, address, postal code, city, website and (optionally) email for any business category in any Danish town. A true Danish business directory scraper and a practical proff.dk alternative for lead generation: type elektriker + Aarhus, press Start, export Denmark B2B leads to JSON, CSV, or Excel.

Both sites are run by the same company on the same platform and hold the same company data, so one actor covers both — pick the directory in the input, or paste start URLs from either.

Business listings only — the actor collects company records, not private persons.

What can you do with it?

• Get Danish business leads with phone numbers — find virksomheder med telefonnummer — every record carries the phone shown on the directory, normalized for dialers and CRMs
• Export krak.dk and degulesider.dk search results to CSV or Excel — hent virksomhedsliste til Excel — one clean row per company, stable columns, ready for import
• Build B2B prospect lists by trade and city — find alle elektrikere i Aarhus — tømrer in Odense, revisor in København, VVS nationwide
• Scrape company websites and addresses — find hjemmeside og adresse på virksomheder — the company's own website is matched straight off the result page, no extra cost
• Enrich leads with email — find e-mailadresse på virksomheder — opt-in mode opens each company's detail page, where the directory keeps the email
• Feed AI agents and pipelines — structured JSON output via API, webhooks, or the Apify MCP server, so agents can call this scraper as a tool

Pricing

$2 per 1,000 business records ($0.002 per record). You pay only for records actually delivered to your dataset — a blocked or empty search costs nothing.

Scenario	Records	Run cost
Electricians in Aarhus	~500	~$1
One trade across the 20 largest Danish cities	~3,000	~$6
Nationwide category sweep	~10,000	~$20

Listing mode vs enriched mode — read this before switching on email

Residential proxy traffic is billed by Apify on your account, separately from the actor price. The two modes are worlds apart there (measured on the sibling eniro.se actor, which runs the identical platform):

Mode	Fields	Proxy data	Approx. proxy cost
Listing (default)	name, phone, address, postal code, city, website, profile URL	~0.017 MB/record	~$0.45 per 1,000 records
Enriched (fetchContactDetails: true)	+ email from each company's detail page	~1.89 MB/record	~$18.75 per 1,000 records

Enriched mode is fully supported but opt-in, and it costs far more in proxy traffic than the actor itself. Most people do not need it: many company websites are already in listing mode, because the directory puts them on the result page. Switch on Fetch email & website only when you specifically need email addresses, and cap maxResultsPerSearch for the first enriched run.

How to scrape krak.dk (step by step)

1. Click Try for free — you need a free Apify account.
2. Leave Directory to scrape on krak.dk, or switch it to degulesider.dk.
3. Enter one or more Search terms in Danish (e.g. elektriker, tømrer, VVS).
4. Enter Locations (e.g. København, Aarhus, Odense) — or leave empty to search all of Denmark.
5. Optionally adjust Max results per search. Leave Fetch email & website off unless you need emails (see the proxy-cost table above — roughly $18.75 per 1,000 records of proxy traffic on your account).
6. Click Start, then open the Dataset tab and export as JSON, CSV, Excel, HTML, or XML.

Prefer pasting URLs? Put krak.dk or degulesider.dk listing pages straight into Start URLs instead of search terms — both hosts can be mixed in one run, and each record records which one it came from.

Input example

{
  "site": "krak.dk",
  "searchTerms": ["elektriker", "tømrer"],
  "locations": ["Aarhus", "Odense"],
  "maxResultsPerSearch": 200,
  "requestDelayMs": 5000,
  "fetchContactDetails": false,
  "includeSoleTraders": false
}

Output example (one dataset record)

{
  "name": "Saabye El",
  "phone": "40000001",
  "email": "kontakt@saabye-el.dk",
  "address": "Bredkildegade 10A",
  "postalCode": "8240",
  "city": "Risskov",
  "categories": ["elektriker"],
  "website": "http://www.saabye-el.dk",
  "profileUrl": "https://www.krak.dk/saabye+el+risskov/65940853/firma",
  "site": "krak.dk",
  "sourceUrl": "https://www.krak.dk/elektriker+aarhus/firmaer",
  "searchTerm": "elektriker",
  "location": "Aarhus",
  "scrapedAt": "2026-08-18T10:30:00+00:00"
}

Missing values are explicit null, so CSV and Excel exports always have the same columns.

Run status and the run SUMMARY

Every run writes a SUMMARY record to the run's key-value store, and the run status follows one rule: a run that delivered rows succeeds. If some searches burned while others delivered, the run is SUCCEEDED with those failures listed in the SUMMARY — so a partial run is never discarded and re-run, and you are never charged twice for the same rows.

SUMMARY field	What it means
recordsPushed	Rows delivered to the dataset and charged — the two are always equal, and this includes rows delivered by a search that burned later on.
failures	One entry per search that burned: the search term, the location, the host and the named error.
failedCombinations	How many searches are in failures.
listingsSeen	Listings the directory showed, before filtering — what makes a thin dataset explainable.
soleTradersFiltered	Of those, dropped as sole traders listed under a personal name (set includeSoleTraders to keep them).
unreadableEntriesSkipped	Of those, dropped because they carried no readable name at all — the schema-break signal.

The run is marked FAILED when it delivered nothing at all — and then the status message says why — and in one other case: if contact enrichment dies platform-wide (a detail-page wall), the run aborts and is failed even though rows were already delivered. In that case the SUMMARY still reports — and you were still charged for — everything delivered up to the abort, so do not re-run it blind: check recordsPushed first.

Use it from code or AI agents

curl -X POST "https://api.apify.com/v2/acts/worktech~krak-scraper/run-sync-get-dataset-items?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"site":"krak.dk","searchTerms":["elektriker"],"locations":["Aarhus"],"maxResultsPerSearch":100}'

The actor also works with n8n, Make, Zapier, webhooks, and the Apify MCP server — its structured output is designed to be consumed by AI agents directly.

Why this scraper?

• Two directories, one actor — krak.dk and degulesider.dk run on the same platform and hold the same records. You do not need two subscriptions, two runs, or two mental models, and every record names the host it came from.
• The URL scheme that actually works — the directory silently ignores ?where= and ?page= query parameters and serves the same national top-25 regardless. This actor uses the path-based scheme (/elektriker+aarhus/firmaer/2) that the site's own rel="canonical" and rel="next" tags declare, so city filtering and pagination really apply.
• Danish company names handled properly — Danish legal forms (ApS, A/S, I/S, K/S) and Danish trade words are both recognised, so real companies like Saabye El and Harlev El are not thrown away as private persons by a business-only filter written for another language. Six of 25 companies on our first captured page were lost that way before this was fixed.
• Pagination that actually stops — this platform advertises a "next page" on every result page, even the last one. The actor ends a search when a page comes back with no companies, instead of grinding through empty pages on your proxy bill.
• Cloudflare-aware by design — anti-detect Firefox (Camoufox) with locale/geo alignment, cookie-consent handling, paced fresh-tab detail fetches, and local exponential backoff on blocks.
• No silent failures — if the directory changes its page structure, the run FAILS with a named error instead of quietly handing you an empty dataset and a proxy bill. A blocked search surfaces as a named error in the run summary instead of quietly returning nothing, and a sustained contact-page wall aborts the run rather than delivering half-empty records. You are never billed for records you didn't receive: data is pushed before it is charged.

Also from worktech: Eniro Scraper for Sweden, Gulesider Scraper for Norway, and Lean Shopify Scraper.

FAQ

What is the difference between krak.dk and degulesider.dk? Branding, mostly. They are run by the same company on the same platform, and the same search returns the same companies with the same company ids on both. Pick either — the actor handles both identically and tags each record with its source.

Is this a proff.dk alternative for lead generation? Yes in outcome: both cover Danish businesses. This actor scrapes the Yellow Pages sites, whose result pages already expose phone, address and (for many companies) the company's own website — which is what makes cheap bulk lead lists possible.

Do I get email addresses? When the directory has one, yes — but only in enriched mode. Email lives on the company detail page, so switch on Fetch email & website (adds roughly $18.75 per 1,000 records of residential-proxy cost on your account). Coverage varies by trade — not every Danish business publishes an email.

Do I need enriched mode to get websites? No. Many company websites come out of the default listing mode at no extra proxy cost. Enriched mode is about email, and it fills in websites the result page did not show.

Can I export the leads to CSV or Excel? Yes. Every run produces a dataset downloadable as CSV, Excel, JSON, HTML, or XML from the Apify Console, or via API.

Does it collect private persons? No. The actor is business-only by policy. Sole-trader businesses listed under a personal name (common for tradespeople) are excluded by default and only included if you switch on includeSoleTraders.

What should I put in Search terms — Danish or English? Danish, matching the directory's own categories: elektriker (electrician), tømrer (carpenter), VVS (plumbing), tandlæge (dentist), revisor (accountant), frisør (hairdresser).

Why is the run slower than a plain HTTP scraper? Both sites sit behind Cloudflare bot protection. The actor uses a real anti-detect browser with human-like pacing — that's what makes the data come out correct and complete. The default listing-only mode is the fast one; enriched mode adds one detail-page load per company.

Is it legal to scrape krak.dk? The actor collects publicly visible business information (the same data any visitor sees) and no private-person data. Whether scraping fits your use case depends on your jurisdiction and the site's terms — when in doubt, consult a lawyer.

Limitations

• Email coverage depends on the directory — companies without a published email simply have email: null, in both modes.
• Requires Danish residential proxies for reliable runs — the default proxy configuration is Apify residential, country DK. Datacenter proxies will get challenged far more often.
• Detail-page enrichment (opt-in) adds one page load per company — expect roughly 3–7 seconds and ~1.89 MB of proxy traffic per enriched record.
• Hyphenated El- names are withheld when ambiguous. El-Installation (a trade compound) and El-Sayed (a family name) look identical to any rule. When the second part is not a recognised trade word the record is treated as a private person and withheld — publishing an individual is a privacy breach, withholding a company is not. Switch on includeSoleTraders to get them.
• Pagination is capped at 50 pages (~1,250 records) per search combination as a runaway guard — split very large categories by city to get everything.
• A search returns companies from a wider area than the town you typed. The directory pads thin result sets with nearby businesses — a 9-result search still fills 25 rows per page and keeps going. Each record carries its own city and postalCode so you can filter after export.
• Running both directories doubles the rows, not the coverage — they hold the same data. Scrape one, unless you specifically want both source URLs.

Status

v0.1 — sibling of the production Eniro Scraper for Sweden, which runs on the same Eniro Group platform. Unit tests pass offline against pages captured live from krak.dk and degulesider.dk, including a real Cloudflare challenge page.