Data sources and crawler policy

Boards we read

Board data comes only from each board's public pages, the ones a signed-out browser can open, and from the new-campaign emails boards send to the accounts we registered with them. We read each board with its permission and take no feeds, webhooks or private data from any of them.

What we read

• Campaign pages. The title, the rate on each platform, the minimum views and cap per clip, the budget total, spent and left, the rules when they are public, and the counts of clippers, posts, approvals and views where the board shows them.
• Board terms. Fees, payout methods, the minimum withdrawal and the minimum age, from each board's public terms and pricing pages, each linked on its board page.
• New-campaign emails. An email only tells us a campaign exists. It never adds one by itself, because we read the campaign's public page first, and we never open the links in it.
• Briefs. The public documents and sites a campaign links, for its rules and its footage.
• Footage. Public source videos a brief names, to find the moments worth clipping and score how clippable they are. Their audio is deleted after 3 days, frames after 7 and transcripts after 30.
• Top clips and example posts. The view counts and links a board shows for its top clips, and the view counts of a brand's example posts. We show those posts as the platform's own embed and never download them.
• Brand pages. A brand's public site, for its name, logo and official links.

How often we read

Each campaign is read on a schedule set by how much it moves. Where a board's list page carries every campaign's budget and rates, as Content Rewards' does, we read that list every 5 minutes and each campaign's own page every 6 hours for what only it shows. Alpha Scores are rebuilt every 10 minutes from the newest reads.

Our crawler

ClippingAlphaBot fetches board pages and briefs, and ClippingAlphaMedia fetches the footage a brief links. Both name their own page in their user agent, read robots.txt before any page on a host, fetch one page at a time per host at least a second apart and back off on a 429 or 503. They send no login of ours, never sign in and never submit a form.

AI crawlers welcome

Our robots.txt lets every crawler read every public page, search, answer and training crawlers alike. It names 26 crawlers with the same rules as everyone else, among them GPTBot, ClaudeBot, PerplexityBot, Google-Extended and CCBot, and its Content-Signal line reads search=yes, ai-input=yes, ai-train=yes. Only private paths, such as accounts, the API and outbound links, and sorted or filtered copies of lists stay closed.

Cite our numbers freely, with a link to the page they come from.

Models we use

Language models from Fireworks AI read public briefs, board listings, transcripts of public campaign videos and public brand pages, to pull out rules and the moments worth clipping. Azure Speech can transcribe public campaign videos. Models never receive account, visitor or alert data.

What we never do

• Sign in to a board, or read anything a signed-out visitor cannot see. Rules a board hides until you join stay hidden on our pages too.
• Take a feed, a webhook or a private API from a board.
• Open the links in a board's email.
• Submit a form or post anywhere.
• Download a clipper's posts. They show as the platform's own embed, loaded when you press play.
• Collect clippers' names or email addresses.

The privacy policy has the full detail on public campaign data and how long we keep it.