Prepared by: Maverick
Overview To ensure campaigns on Reach.cat are safe from bots, fake engagement, and manipulation, we’re introducing a two-level trust scoring system:
Creator Score (Macro): Checks if the creator’s account is authentic.
Content Score (Micro): Checks if the post itself looks suspicious.
1. Creator Score (Macro) This score gives an overview of how trustworthy a creator is based on their account history.
Key Checks:
Account Age: Older accounts are more reliable.Example: An account that’s over a year old gets a higher score than one made last week.
Engagement Ratios: Instagram: Like-to-comment ratio should be around 0.35 or higher. Twitter: Comment view-to-post view ratio should be 5% or higher. Example: If a post has 10k views, the comments under it should have at least 500 views.
Follower Growth: Sudden growth with no engagement likely = fake.
Consistent Posting: Accounts that post regularly look more legit.Example: At least 1 or 2 posts a week for 3 months.
Quote Tweet Spam: On Twitter, check if the creator replies to a lot of unrelated viral tweets. This is common for fake boosting.
2. Content Score (Micro) This score focuses on each piece of content submitted.
Red Flags: High Views, No Engagement: Example: A tweet with 10,000 views and 3 comments is most likely botted.
Suspicious Comment Patterns: Example: Dozens of emoji-only comments or “Great post!” repeated 10 times.
Reply Farming: Posting campaign links under viral tweets to steal views.
View Distribution: Organic posts = views on post + some views on comments. If comment views are super low, it’s suspicious.
Open Source & SaaS Detection Tools We can borrow ideas and even plug into tools already built to catch bots. I listed some tools here:
👉 Botometer (Twitter) Rates how likely an account is a bot based on over 1,000 signals. Use it to score Twitter creators before approving them. https://botometer.osome.iu.edu/
👉 BotD by FingerprintJS Detects non-human browser behavior. Can help protect Reach.cat from fake sign-ups. https://github.com/fingerprintjs/BotD
👉 SEON Real-time fraud scoring across email, device, IP. Could be used to flag creators at signup or campaign submission. https://seon.io/
Some General Prevention Ideas
Join Campaign = Review First: When someone clicks "Join Campaign," we check their score first.
Only High Trust = Access to Premium Campaigns
Live Monitoring: Update scores if a creator posts botted content later on. Example you bot twice and your score is reduced drastically.
Guide Advertisers: Teach brands what fake engagement looks like.
Set clear score cutoffs for certain campaigns (e.g., must have 70/100 to join)
Build a dashboard showing creator & post scores
Keep testing and improving these rules as we get more data because botters and farmers evolve so we must.
Detailed Specifications for Creator/Content Scores: