DEV Community

Cover image for Ad verification proxies: checking ads market by market
Andrew
Andrew

Posted on Originally published at proxymint.com

Ad verification proxies: checking ads market by market

Originally published on proxymint.

How to verify ad placements, creatives and landing redirect chains per market from a consumer origin, which proxy tier fits, and how to size a setup.

TL;DR

  • Verification needs three things. An exit the target places inside the campaign's market, a consumer network rather than a hosting one, and one origin held through the whole redirect chain.
  • Platform tools come first. Google's Ad preview and diagnosis tool, the Ads Transparency Center and the Meta Ad Library show an advertiser's own ads by location for free. A proxy check starts where they stop: third-party placements and the landing page a visitor actually gets.
  • Hosting IPs see a different ad world. The MRC's 2020 invalid traffic standard has accredited measurers filter invalid data-center traffic from AWS, Google and Microsoft, and cloakers key on hosting networks too.
  • Log the chain hop by hop. Follow 301, 302, 303, 307 and 308 redirects manually, record the exit IP at start and end, and rerun client-side hops in a headless browser.
  • Match the origin to the audience. Residential sticky sessions for desktop and home placements, mobile exits for carrier campaigns, one fixed ISP IP per market for daily audits.

Key takeaways

  • Record exit facts with every check. Log the exit IP, its country, city and network type at the start and end of each chain so proxy errors are not mistaken for ad problems.
  • Use a session longer than the chain. A 10 to 30 minute sticky session covers any redirect chain; per-request rotation breaks it.
  • Find cloaking in the diff. Run the same URL from a consumer exit and a hosting exit in one market; a different final page on the hosting side is the signature.
  • Trace only URLs the team owns or audits. Clicking other advertisers' paid ads charges them and pollutes their data.
  • Price the job by what it captures. Header-only chain traces are light; full-page screenshots drive the GB bill, so measure one day before buying a package.

Read the full guide on proxymint, with the commands, tables and sources.

Top comments (0)