YouTube Channel Email Scraper - Business Contacts

Give it a list of YouTube channels. It returns the business email each channel publishes itself, and says where each address came from. When a channel publishes no email but links a website, the site and its contact page are checked too.

Nothing here comes from behind YouTube's "view email address" button - that asks a human to pass a check, and this Actor does not touch it. Only what a channel puts in public.

What you actually get, measured

30 channels were checked on 2026-09-02. The channel list, the per-channel result and the script that produced them ship with this Actor (test/survey-channels.mjs, test/_channel-survey.json), so the number below can be reproduced or challenged. All 30 pages were readable.

	Channels	Share
Email published in the channel description	6	20%
No email, but a website is linked	7	23%
Neither	17	57%

So fewer than half of channels are reachable - directly, or through the site they link. The rest publish nothing, and no scraper can change that. This one tells you which group a channel is in instead of returning a blank row that could mean anything.

One row per channel

Field	Example
channelId	UCBJycsmduvYEL83R_U4JriQ
title	Marques Brownlee
emails	["business@mkbhd.com"]
emailSources	["channel description"] or ["https://example.com/contact"]
websites	["https://mkbhd.com"]
contactPagesCrawled	how many pages of that site were opened
ok	true only when at least one address was found
reason	why the row is empty, when it is

emailSources is the part most tools leave out. An address found in the channel's own description and an address found on some page of a linked site are not the same quality of lead, and you should be able to tell them apart before you write to anyone.

The channel you asked for, not the one next to it

A YouTube channel page carries data for other channels too - recommendations, shelves, related creators. Reading "the first channel id on the page" gives you the wrong channel, and the output looks perfectly normal while being wrong. (We know because we did exactly that in a different project on 2026-08-29 and published numbers for the wrong channels.)

This Actor reads only the page's own channelMetadataRenderer block, and a test feeds it a page with a second channel embedded to prove the wrong one is never picked.

Input

{
  "channels": ["@mkbhd", "https://www.youtube.com/@LinusTechTips", "UCYO_jab_esuFRV4b17AJtAw"],
  "crawlWebsite": true,
  "maxPagesPerSite": 3,
  "respectRobots": true
}

Handles, channel URLs and UC… ids all work and can be mixed. Anything unreadable comes back as a row saying so - never as a silent skip.

Field	Default	Meaning
channels	(required)	What to look up
crawlWebsite	true	Also open the website a channel links, when it publishes no email
maxPagesPerSite	3	How many pages of that site to open before giving up. Stops at the first address
respectRobots	true	Skip what the site disallows
timeoutSecs	20	Per page

Pricing

Charged only for channels where an address was actually found. Channels that publish nothing, channels that could not be read, and inputs that were not channels are all delivered with a reason and cost nothing.

That matters here more than in most Actors: on the list above, 57% of channels published no contact at all. Some Actors in this category bill per input row and others, like this one, bill per contact found - worth checking which one you are buying, because on a list like that the difference is more than double.

Checks

120 automated checks, and 16 deliberate defects in the shared extraction layer plus 13 more aimed at the channel parser - 29 in total. Every defect is inserted on purpose to confirm a check turns red: picking the wrong channel's data, accepting a broken metadata block, treating social links as the creator's website, marking an empty row as ok, dropping the source of an address, letting a no-reply or placeholder address count as a contact, stamping rows with the current time instead of the run time. All 29 are caught (npm test, npm run mutate).

Other tools by NeverEmpty

• youtube-transcript-reliable - transcripts, with the failure reason when there is none
• youtube-channel-transcripts - the same engine across a whole channel
• website-contacts - emails, phones and socials from a list of websites