YouTube Sponsorship & Brand Deal Researcher

Overview

This Actor collects recent videos from the public YouTube Atom feed for channel IDs you supply. It returns video titles, public descriptions, links, dates and view counts where the feed provides them. It flags explicit sponsorship disclosures, possible affiliate links and labelled discount codes for human review. It does not claim that a brand deal exists.

Input

Provide channelIds as an array of UC channel IDs or canonical HTTPS /channel/ URLs. Up to 50 inputs are accepted; duplicates are removed. @handles, arbitrary websites, credentials in URLs and local addresses are rejected before any request. maxItems caps returned rows; maxVideosPerChannel caps examined entries per channel. onlyWithSignals removes rows without a detected disclosure, possible affiliate link or labelled code.

Example

The reproducible sample input is in examples/input.json. Run npm install followed by npm test, then use the existing factory release tool with --local-run --write-example. examples/output.json is populated only from a real successful run, never from the synthetic test fixture.

Output

Each row contains videoId, channelId, channelName, title, url, publishedAt, updatedAt, description, views, thumbnailUrl, sponsorshipStatus, sponsorEvidence, affiliateLinks, discountCodes, externalLinks, source and checkedAt. Missing feed fields remain null or empty rather than being invented. sponsorshipStatus is unverified-signals or no-explicit-signal. Neither value proves that a commercial relationship does or does not exist. Evidence is quoted from the public description.

The default key-value store also contains RESEARCH_SUMMARY and REPORT, including failed channels. HTTP failures and invalid feed responses are not emitted as video rows. If all channels fail, the Actor fails rather than returning a paid placeholder. A legitimately empty feed can succeed with zero rows. Duplicate video IDs are removed globally.

Pricing

Pricing is $0.005 per returned video (video-analyzed), equivalent to $5 per 1,000 videos, plus $0.00005 for the Actor start event at the default memory setting. The synthetic default-dataset event is not charged in addition to the video event. Apify Console is authoritative for current pricing and any memory-based start-event multiplier. No paid proxies, external AI APIs or YouTube API key are used. Local test execution does not purchase credits. Use the existing release.mjs usage guard before cloud runs.

Limitations

This is recent-feed research, not a complete YouTube archive, keyword search engine, video downloader or transcript scraper. The provider controls which recent entries are present. Views and descriptions can be absent or stale. Sponsorship matching is heuristic and recognizes selected English, Polish and German disclosure phrases; it can miss disclosures, and links or codes can be non-commercial. No external links are opened, no user account is logged in, and no CAPTCHA, access control or paywall is bypassed. Review evidence yourself before business decisions.

Verification

Regression tests cover real Atom structure, XML entities, output fields, missing and unsafe inputs, per-channel and global caps, duplicate channels/videos, failed and malformed feeds, empty feeds, partial success and filtering. Test data is synthetic and isolated from production outputs. Run receipts distinguish local execution from cloud execution and public publication.

Protocol reference: https://developers.google.com/youtube/data/guides/push_notifications