Find a company's website from its name (bulk): 96% on known companies, $0.80 per 1,000 found

By Keencrest · 2026-11-01

You have a column of company names, from a trade-show list, a CRM export or an award list, and no website next to them. Most enrichment and email tools want a domain, not a name. This guide shows how to turn the names into domains in one run, what share you can expect to get right (it depends heavily on how well known the companies are), and what it costs.

Price

Job	Cost
1 website found	$0.0008
1,000 websites found	$0.80
A name that is not found	$0, returned with the best guesses

You pay per website found at or above your confidence floor (default 0.5). There is no start fee and no monthly plan; the Apify free plan includes monthly credit, and a maximum spend per run stops the Actor when it is reached.

How it works

For each name the Actor asks open data first: Wikidata (CC0, through the Wikimedia API) states the official website of many companies, and for Norwegian and Finnish companies the national business registers (Brreg, PRH) are asked for the website the company registered. It also tries the domains the name spells, such as bluebottlecoffee.com for Blue Bottle Coffee. Then it opens every candidate and keeps it only if the company's own home page carries the company's name; parked and for-sale pages are rejected and redirects are followed to the final domain. Each row has a confidence score, the method that found it and the evidence in plain words, plus the logo URL and LinkedIn company page when the site links them. It reads no search engine.

Measured result 1: 50 well-known companies

On 2 November 2026 we ran 50 companies of mixed size from 20 countries on Apify (run cf0APjxPKKDPbhi8c, build 0.7.6; the first measurement on build 0.6.1, run 7nrHtP4mi552Ynb2n, gave the same counts). The input was just names, some with a country:

Company size	Names	Right website	Not found
Large	15	15	0
Mid-size	20	20	0
Small	15	13	2
All	50	48 (96%)	2

The run took 87 seconds. Wikidata named 46 of the 48 correct sites; the other 2 came from the name's own domain. Without the open-data step, the same 50 names gave 33 right (66%) in a local test. The 48 found websites would have cost $0.04 and the 2 misses nothing. The two names it missed are companies whose legal name differs from their brand; searching them by brand name is the fix. Here "correct" means the returned domain is one of the company's own official domains.

Measured result 2: small local businesses (the limit)

The harder case is a list of local firms: plumbers, salons and cafes taken from Google Maps leads, and companies from job-board lists. We ran 98 such names (run fuTBHJZjM3nhh6Grq, same build and date; the 0.6.1 run LbcnwJsGRWIfks7AH gave identical counts). It took 132 seconds. For 59 of them we knew the real website. Result on those 59:

Outcome	Names	Share of 59
Right website	23	39%
Wrong website	5	8%
Not found (free)	30	51%

The scorer's tally covers 58 of the 59: the last row, a plumber, came back with a different site from the one we had on record and needs a human look, so it is in no bucket. Read the table as "about 4 in 10 right, about 1 in 10 wrong, about half not found". Of the 28 rows it returned as found, 23 were right (82%). Two of the five wrong ones were the right company's site in another country, for example the US site for a London branch of the same brand. A small local business without a Wikidata entry is found only if its own name is also its domain and its home page says so, which is why the rate is so much lower.

If you work on local lists, set minConfidence to 0.8 so a wrong site costs you less than a miss, add the city to each name (Joe's Pizza, New York), and expect to find the rest by hand or by another route. If your list has a website, phone or postalCode column (Google Maps lead rows do), pass it as a dataset: the page is checked for the phone or postcode, which raises confidence.

Steps

Open the Actor on Apify and press Try for free.
Paste the names into companies, one per line, with a country or city after a comma where you have it. Or upload a spreadsheet as a dataset and pass datasetId.
Run, then filter on found = true and sort by confidence. Rows with found: false carry the best guesses in alternatives.
To get emails next, run the list walkthrough on the dataset, or set findContacts to true to do both in one run. The names-to-emails guide shows a measured 20-name run.

Limits to know

Long-tail local firms: about 4 in 10 found, as measured above. This is a tool for lists of known companies first.
Common names with no location hint can match a same-named company. Add a country, city or industry.
Speed: the open-data lookups run one at a time under the Wikimedia API rules, so 1,000 names take roughly 30 to 40 minutes.
It returns one main site per company; subsidiaries and country sites appear in alternatives. Brand-new companies with no web presence are not found.
Figures were measured on build 0.7.6, the live build when we re-ran both sets; the counts match the first measurement on 0.6.1. Check the Actor's README for the current numbers.
You are responsible for how you use the results, including marketing and data-protection law where you operate. The Actor returns company websites, not people's data.

Try it: Company Domain Finder on Apify. $0.0008 per website found; names not found are free. The Apify free plan includes monthly credit, so a test run costs nothing out of pocket.

Related guides

All Keencrest guides