IdeaWell

Project ideas from Hacker News discussions.

I spent $220 on Google app ads and 60% of the installs were robots

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

Theme 1: Bot‑driven ad fraud is widespread

"this is unfortunately very common - I kept running very small ads regularly over many years (~10 years by now) and the % of bots has been steadily increasing, and Google doesn't really have any system to report those reliably to them." — yalok

Theme 2: Platforms have little incentive to fix fraud

"Taking a blind guess because I have never worked at Google, but I would assume there is one organization that has the data, and another organization that can block IPs from clicking on the ads and consuming the spend. There is a byzantine process preventing that second org from getting the data along with a lack of motivation because it would decrease ad‑spend, which is one of their key metrics." — n0us

Theme 3: Fraudulent ecosystem profits bot farms at advertiser expense

"Bot farm is operated by a company that displays Google ads. This bot farm fakes traffic to the bot farm owner and simulates conversion Google wants to see. Google sees the high conversion and happily serves more ads to the bot farm owner. Bot farm owner is paid by Google for displaying the ads. The person paying for the ad gets screwed." — wilbo

Theme 4: Advertisers can mitigate via IP exclusions and tight campaign controls

"If you have their IP addresses in your dashboard, go to Google Ads > Admin > Account Settings > IP Exclusions. Then, add the entire network/data center range there (i.e. 123.4.5.). In 99% of cases, these bot networks are not run from residential providers." — phenomen*

🚀 Project Ideas

Generating project ideas…

FraudIP DB

Summary

  • A crowdsourced, continuously updated database of IP addresses and networks known to generate fraudulent ad clicks and installs.
  • Provides an API for ad platforms and advertisers to automatically exclude these IPs from campaigns, reducing wasted ad spend.

Details

Key Value
Target Audience Mobile app developers, performance marketers, ad agencies running Google/Meta ads
Core Feature Real-time IP lookup API + downloadable blocklists (CSV/JSON) of fraudulent IP ranges
Tech Stack Backend: Python/FastAPI, PostgreSQL, Redis for caching; Frontend: React for admin UI; Deployment: Docker/Kubernetes
Difficulty Medium
Monetization Revenue-ready: Subscription tiered by API calls (free tier up to 10k lookups/mo, paid plans from $49/mo)
#### Notes
- HN users asked "Is there an open database of this?" and noted maintaining personal exclusion lists of over 4000 networks; a shared DB would save effort. (yalok: "Google doesn't really have any system to report those reliably to them.")
- Enables community-driven fraud prevention; could spark discussion on IP reputation and evasion tactics.

BotBehavior SDK

Summary

  • A lightweight mobile SDK that logs install events and post-install user interactions to detect bot-like behavior (zero-time installs, lack of engagement, abnormal click patterns).
  • Sends real-time alerts and provides a dashboard for developers to identify and block fraudulent installs before they pollute ad metrics.

Details

Key Value
Target Audience Indie game developers, app publishers using ad networks (Google AdMob, Meta Audience Network)
Core Feature On-device heuristics + optional cloud scoring to flag suspicious installs (e.g., installs with <5 sec session, no UI interaction)
Tech Stack Native SDKs (Android Java/Kotlin, iOS Swift), backend: Node.js/Express, MongoDB, optional TensorFlow Lite for on-device model
Difficulty Medium
Monetization Revenue-ready: Freemium – free SDK with basic logging; premium analytics ($29/mo per app) for advanced scoring and alerting
#### Notes
- Commenters noted bots that install then do nothing (yalok: "They started installing the app. Then running the app once and do nothing in it.") and zero-time installs without clicks (nickabe: "29 of the 30 zero-time installs had no click").
- Provides concrete data to dispute fraudulent charges; HN audience appreciates open-source tools that improve ad ROI.

AdRefund Helper

Summary

  • A SaaS tool that aggregates campaign data (clicks, installs, post-install events) from Google Ads, Meta Ads, and MMPs, automatically builds evidence packages for suspected fraud, and facilitates refund/dispute submissions to ad networks.
  • Reduces manual effort and increases success rate of recovering wasted ad spend.

Details

Key Value
Target Audience Performance marketers, small to mid-sized businesses spending >$5k/mo on paid user acquisition
Core Feature One-click fraud report generation: exports CSV/PDF with IP lists, timestamps, behavioral anomalies, and pre-filled dispute forms for Google/Meta
Tech Stack Backend: Ruby on Rails or Python/Django; Integrations: Google Ads API, Meta Marketing API, Firebase/Adjust webhooks; Frontend: Vue.js
Difficulty High (due to API complexity and varying dispute processes)
Monetization Revenue-ready: Percentage of recovered funds (e.g., 15% of refunded amount) or flat monthly fee from $99/mo
#### Notes
- Users expressed frustration with getting refunds: "I’ll report back on the refund." "Good luck on that." "I'm not holding my breath, but I'll report back either way." (nickabe)
- Automating evidence collection addresses the lack of a reliable reporting system mentioned by yalok: "Google doesn't really have any system to report those reliably to them."
- Could generate HN discussion on best practices for fraud disputes and network transparency.

ProxyShield

Summary

  • A detection service that identifies traffic originating from known residential proxy and VPN networks, providing real-time scoring and blocklist options for ad campaigns.
  • Helps advertisers filter out low‑quality clicks from proxy farms used to fake residential IPs and evade basic IP bans.

Details

Key Value
Target Audience Advertisers concerned about sophisticated fraud (e.g., those seeing spikes in installs from residential‑like IPs)
Core Feature IP reputation API that returns probability of being a residential proxy/VPN, updated via honeypot sensors and public proxy lists
Tech Stack Backend: Go for high-performance IP lookups; data pipeline: Apache Kafka + Spark to process proxy‑list feeds; Frontend: React monitoring dashboard
Difficulty High
Monetization Revenue-ready: Usage‑based pricing ($0.001 per 1k lookups) with minimum monthly commitment $79
#### Notes
- Discussion highlighted residential proxies being bought for fraud and scraping (walrus01: "residential proxies for sale"; "pay for these things is 'access to 500+ TB Jellyfin-like movie services'"). Users noted fraudsters evolve to bypass simple IP bans.
- Provides a proactive layer beyond basic IP exclusions; HN community interested in networking, proxy detection, and ad‑tech integrity.
- Potential for open‑source proxy list contributions, fostering transparency.

Read Later