Company Financials Scraper — Valuation & M&A (Craft.co)
Pricing
from $4.00 / 1,000 results
Scrape company financial intelligence from Craft.co — revenue, income statements, private valuations, market cap, M&A, executives & competitors as clean JSON. Low-cost Crunchbase / PitchBook alternative for VC/PE deal sourcing, M&A research, sales prospecting & CRM enrichment.
FAILED_PAGES. When at least one failure is on our side — Craft.co's bot protection on every attempt, an error, a requested company the run never reached — the run still ends FAILED, so a real breakage stays visible. Such a run is the user's to fix, not a broken Actor; in the 30 days to 2026-10-10, 8 of the 37 runs by users ended FAILED.run['defaultDatasetId']. With the current Apify client for Python (version 3) call() returns a run object, not a dict, and that line raised TypeError: 'Run' object is not subscriptable — after the run itself had finished. The example now reads run.default_dataset_id; with client version 1 or 2, keep run['defaultDatasetId']. Checked with apify-client 3.2.1, 2.5.1 and 1.12.2.Stripe, Airbnb, Notion Labs, depth 2) went from 12 and 26 rows with 1 of 3 requested companies to 93 rows with 3 of 3 (three runs); a single-company run failed 3 times out of 6 before and 0 out of 6 after. On the platform the country default alone (0.4.13 with the country set in the input) gave 77 rows and 3 of 3 against 8 rows and 1 of 3.FAILED_PAGES entries carry category: "not-found" (no profile at the address(es) tried) or "blocked" (bot protection on every attempt — run again) instead of the single "not-found-or-blocked". Neither is billed — the lookup-fee terms announced for 17 October 2026 are unchanged.url and slug. This also works when you ask by name: OpenAI is delivered as Craft.co's current profile "OpenAI Foundation", because craft.co itself redirected the address and the name you typed leads the new name (before, it was reported as "resolved to another company"). A rename to a differently named company (Square → Block) is still not delivered for a name — use the craft.co URL — and such an entry is marked viaRename and will not be charged the lookup fee.FAILED_PAGES (not-reached) instead of being dropped silently.FAILED_PAGES when that second fetch fails.openai → openai-foundation, snowflake → snowflake-computing, scale-ai → scale. Removed from the AI / ML preset because craft.co has no profile at these addresses: cohere, midjourney, google-deepmind.company-lookup, and a delivered record drops from $0.004 to $0.0035 — a delivered company costs $0.004 as before. A company you asked for that resolves to another company or has no Craft.co profile costs only the $0.0005 lookup (a valid company that stays blocked after every retry is indistinguishable and is charged the same). Lines never looked up, preset anchors and discovered competitors that fail, and other errors on our side stay free. The code is live now and charges nothing extra until the new price takes effect (src/billing.js).Stripe Inc → craft.co/stripe; Goldman Sachs Group → goldman-sachs-group, then goldman-sachs; Procter & Gamble → procter-gamble) and a profile is delivered only if its name matches the name you typed (legal-form, group and generic words such as Inc, Holdings or Technologies are ignored). Before, whatever company sat at craft.co/ was delivered and billed — Meta returned "Meta View", a 33-person company. Such entries now go to FAILED_PAGES and are not billed. Side effect: a renamed company is no longer returned for its old name (Square → Block); use its Craft.co URL.stripe.com or https://www.stripe.com/… → slug candidates, delivered only if the profile's website is on the same domain (craft.co/checkout is a different company at checkout.fi, so checkout.com resolves to craft.co/checkout-com). Before, a domain became craft.co/stripe.com and the run failed. Companies whose Craft.co website is another domain (PayPal → investor.pypl.com, Zoom → zoom.com) do not match by domain — use the Craft.co URL.craft.co/x, http://, www., sub-pages (/stripe/financials → the full profile), query strings, trailing slashes. Before, sub-pages returned partial rows and the other forms failed.companyUrls are read (one company per line). Before, the file was ignored and the run failed with "No companies".matchedBy (url | domain | name) on every row.Notion Labs (a bare Notion is a different company). Proxy wording now matches measurements.src/resolve.js (pure functions) with network-free unit tests in test/ (npm test).queries and maxResultsPerQuery — keys that do not exist in this Actor's input schema — with a placeholder value, and the input table listed those same phantom fields. Anyone who copied the AI-agent, cURL, JavaScript or Python example got a failing run. Every example now uses the real schema and matches the Console prefill: {"companyNames":["Stripe","Airbnb","Notion"],"maxResults":100}input_schema.json, so it lists the fields the Actor actually accepts.query", and industry described as a NACE code.